Past Midnight

Feeling safer online with Firefox

2017-01-13T14:50:00.000+02:00

The latest privacy and security improvements in Firefox

[This post originally appeared on Medium]

Firefox is the only browser that answers only to you, our users; so all of us who work on Firefox spend a lot of effort making your browsing experience more private and secure. We update Firefox every 6 weeks, and every new change ships to you as fast as we can make and verify it. For a few releases now, we have been landing bits and pieces of a broader set of privacy and security changes. This post will outline the big picture of all these changes.

Site Identity and Permissions Panel

The main project involved improvements to the way Firefox handles permission requests from sites that want to do things that the web platform doesn't allow by default - like accessing your laptop's camera or GPS sensor. To find out how our existing model fares, we ran it through a number of user studies and gathered feedback from users and web developers alike.

What we found was clear: users were having trouble making full use of web permissions. Here are some of the observations:

It’s easy by design to dismiss a permission prompt, to prevent websites from nagging you. But it’s not so obvious how to get back to an inadvertently dismissed prompt, which users found confusing.
Managing the permissions of an individual site was hard, due to the multitude of presented options.
It was cumbersome to grant access to screen sharing. This was because it was difficult to select which area of the screen would be shared and because screen sharing was only permitted on websites included in a manually curated list.

In order for the open web platform to be on par with the capabilities of native, closed platforms, we needed to fix these issues. So we first focused on putting all privacy and security related information in one place. We call it the Site Identity and Permissions Panel, or more affectionately, the Control Center™.

The Site Identity panel appears when you click on the circled “i” icon – “i” for “information” – on the left side of the Awesome Bar. The panel is designed to be the one-stop shop for all security and privacy information specific to the site you’re on. This includes encrypted connections certificate, mixed content warning messages, tracking protection status, as well as non-default permissions. We were happy to see Chrome adopt a similar UI, too.

Elevated Privileges for Non-Default Permissions

By default, web sites need an elevated level of privilege to access your computer hardware like camera, microphone, GPS or other sensors. When a site requests such a permission and the user grants it, the Site Identity panel will display the allowed item along with an "x" button to revert it. In the case of particularly privacy-sensitive permissions, like microphone or camera access, the icon will have a red hue and a gentle animation to draw attention.

When a site has been granted elevated privileges, the "i" icon in the URL bar is badged with a dot that indicates the additional information present in the Site Identity panel. This lets you assess the security properties of your current session with a quick glance at the awesomebar, where the "i" and lock icons are displayed together.

Users who want even more fine-grained control over all available permissions can go to the Permissions tab in the Page Info dialog (right arrow in the Identity panel -> More Information).

Permission Prompt and Dialog

Permission dialogs are now more consistent than before, both in terms of available actions and messaging.

When a site asks for a permission, a permission prompt appears with a message and iconography specific to the type of permission being requested and the available actions. Most of the time, there will only be two: allow or don’t allow access. The default action will stand out in a blue highlight, making the common action easier to perform.

In the few cases of permission prompts with more than two actions, a drop-down menu will appear next to the secondary action button.

Permanently allowing or rejecting a permission for a site is done by checking the always present "Remember this decision" option.

We have received a lot of feedback about how these prompts are easy to dismiss and how users often couldn't figure out how to bring them back. In the new design, permission prompts stay up even when you interact with the page. You can of course ignore it and continue to use the page normally. But thanks to the persistence of the prompt, it’s now easier to associate site misbehavior – webcams that don’t work, locations that won’t display – with an allow/don’t allow button that needs your response.

Furthermore, disallowed permission requests are now displayed as strikethrough icons in the Awesome Bar to hint at the potential cause of site breakage. For example a video conferencing site will probably not be functioning very well if you reject its camera permission request. So the crossed-out camera icon will remain afterwards, next to the "i" icon, to remind you of that fact.

Going to a different tab will hide the prompt (because it’s specific to the site you have open on each tab), but when the prior tab is selected again, the prompt will reappear.

Audio, Video and Screen Sharing Permissions

WebRTC-related permissions have even more new changes.

For starters, screen sharing now doesn't require sites to be added to a separate whitelist. This means that all sites can now use WebRTC screen sharing in Firefox.

Also, screen sharing now includes a preview of the content that will be shared to make it easier to identify the right screen, application or window to share.

In the riskiest of cases, such as sharing the entire screen or sharing the Firefox application, a scary warning message is displayed to ensure you know what you are about to do.

Moreover, when you have granted a video conferencing site access to both your camera and microphone, reverting the permission grant for one permission will also revert it for the other. This will help you avoid accidentally leaking your private data.

Add-on Panel Improvements

While working on these security improvements we fixed some old platform panel bugs that used to affect all kinds of panels, including those created by add-ons. Therefore if you are using an add-on that displays popup panels you should have an improved experience even if the panels are not related to permission prompts.

Error Pages

And finally, error pages also received some new smarts.

The most common cause for secure connection errors turns out to be user systems having the wrong time. Firefox will now detect when your clock seems way off and will suggest in the error message how to fix it.

Another common cause for broken connections is the presence of a captive portal. Firefox will now detect that case and prompt you to log in the captive portal. Even though some operating systems have built-in support for detecting captive portals, if you regularly use social network accounts to log in, the experience with Firefox will be smoother. This change is now in Nightly and Developer Edition versions and should ship soon in the stable release.

Looking back at what we managed to accomplish in the last few months makes me proud to work with this fabulous team of talented and passionate engineers, designers, user researchers, QA engineers, product and project managers. But of course we are far from being done with privacy and security improvements for our users. Stay tuned for more exciting Firefox privacy and security updates in 2017!

[Many thanks to Bram Pitoyo, Nihanth Subramanya, Tanvi Vyas, Peter Dolanjski, Florian Quèze, and Johann Hofmann for reviewing drafts of this post.]

Valence roadmap

2014-12-30T18:14:00.000+02:00

A few days ago I sent the following note to the Firefox devtools mailing list. The point was to start a discussion about where we want to take Valence (née Firefox Developer Tools Adapter) in 2015. I am reproducing the message below in an effort to broaden the audience, but I would encourage anyone who wants to participate in the discussion to post to the mailing list. However, if that sounds too intimidating (it really shouldn't!), I'm always happy to respond to comments in this blog.

Hey fellow devtoolers!

I know that eggnog and Christmas trees is all that everyone is thinking about right now, but I'd like to take a minute before we all indulge in our holiday festivities, to talk about Valence. And 2015.

2015 will be the year when Valence becomes something more than a promising experiment. It is the year that will (hopefully) see a version 1.0 that has feature-parity with Firefox, for features where the Safari/Chrome debugging protocol aligns well with the Firefox RDP. This is a proposed roadmap of how we'll get there from where we are now.

There are two main themes in this roadmap: do more and do better, and I think we should address them in reverse order. Do-things-better contains what I think we need for cutting a 1.0 release:

Improve existing features (do better)

Get the bundled ios_webkit_debug_proxy working on Windows & Linux
Implement support for the network panel
Handle format specifiers in console.log and friends (%s, %c, etc.)
Display inferred names for functions
Evaluate expressions in the selected frame when execution is paused
Support source-mapped CSS
Break on DOM events
Fix some longstring-related bugs that we currently work around by using plain strings
Fix every XXX and TODO comment in the code
Implement every todoMethod in the actors

Most of these tasks have a well understood scope, so it is just a "small" matter of time and available resources. However, another angle on the do-better theme is not to just fix known problems, but also to make them stay fixed and provide a reliable experience to the user. These tasks belong in that category and should probably be done in parallel with the above:

Improve reliability (do better for realz)

Deploy some testing infrastructure (probably based on ted's work)
Port existing Firefox devtools tests and/or write new ones
Fix stability bugs in Valence

I believe the above lists should be enough for a v1.0, but there are many more things we could do to provide value to our users. Here are some obvious features we could add for a v2.0:

Expand our feature set (do more)

Support for the performance panel (profiler and timeline)
Support for the storage panel
Support for the canvas debugger
Support for debugging embedded web views on iOS and Android, which would let us become a sort of unofficial development environment for Cordova/PhoneGap
Support Firefox Developer Tools in the pipeline (memory profiler, worker debugger)

Not all of these tasks have been scoped or had a feasibility assessment, but there are some indications that we should be able to pull them off with some effort. These features would make the value proposition of Valence and (Firefox Developer Edition by extension) really enticing for many more developers.

But when you have all this, why stop there? The adaptability of Firefox Developer Tools that Valence affords can take us to a whole new level of debugging functionality that developers have always dreamed of:

Full-stack debugging (do way more)

Support debugging server-side node.js applications, with console, scratchpad, debugger and profiler

(Adding async stack support in Firefox Developer Tools would let us combine server- and client-side stack traces for debugging ecstasy)

Research the framework ecosystems of languages that compile to JS and can run on node.js and assess the feasibility of supporting them directly via source maps
Implement adapters for JDWP (to debug Java) and Xdebug (to debug PHP)

Only some preliminary research has gone into these tasks, so their feasibility and effort required is still TBD. Most of the above are further out from a priority standpoint and some may well be in the pie-in-the-sky domain, but I think we should at least investigate node.js debugging in 2015. It would be an exciting engineering accomplishment and could provide a concrete value proposition for Valence and Firefox Developer Edition.

Prioritized shortlist
Given all of the above, and taking difficulty and usefulness into account, I think a reasonable, prioritized, shortlist of Valence tasks for 2015 would be the following:

Get the bundled ios_webkit_debug_proxy working on Windows & Linux
Display inferred names for functions
Implement support for the network panel
Fix stability bugs in Valence
Deploy some testing infrastructure
Port existing Firefox devtools tests and/or write new ones
Support debugging server-side node.js applications
Support for debugging embedded web views on iOS and Android
Handle format specifiers in console.log and friends (%s, %c, etc.)
Evaluate expressions in the selected frame when execution is paused
Support source-mapped CSS
Break on DOM events

Of course, depending on broader organizational priorities, this list may be too short or not much of a shortlist, for an entire year. Not to mention that planning that far ahead is usually futile and an exercise in frustration. Still, this should hopefully provide a useful base for a discussion about what we want Valence to be and how we want to go about it.

Any and all feedback is welcome (before or while you enjoy your eggnog)!

Cheers,
Panos

My 3 year mozversary

2014-04-28T13:39:00.000+02:00

A couple of days ago I started my 4th year at Mozilla and as with all anniversaries, it was a good opportunity to step back and reflect on my accomplishments during this past year and where I'm going next. And since Mozilla practices complete transparency, it also seemed like a good opportunity to sit down and write a blog post, as I never seem to get enough time nowadays to do so. Besides, what's the point of having a blog if not for embarrassing yourself every once in a while?

For a year with 4 different CEOs, the first thought that came to mind regarding my performance was: "a bunch of code, a ton of reviews". I didn't know if that was actually true though, so I set out to collect some evidence. All the code metrics I've come across must always be taken with a grain of salt, so I didn't worry about being completely accurate and I don't intend to gauge my progress on these metrics alone, but at least I wanted some quantification of the kind of impact that I had. Besides refactoring and bug fixing, my most useful feature work in this period includes break on DOM events, closure inspection, the memory actor and improved debugger behavior after a reload (edit: forgot to add the tracing debugger and function re-evaluation in Scratchpad).

The first interesting observation [1] is that I landed 63 patches last year, down from 139 the previous year, which was way bigger than the 56 patches the year before. This was somewhat disappointing and I believe that partly it was due to the initial debugger landing the year before and general work on an immature code base at that time, which made progress much easier to happen in small chunks. I find that I don't usually land separate patches for the same bug, so at least the number of commits is closer to the number of distinct bug reports or new features, than just individual code changes. Another reason for the decrease is that I spend significant time prototyping things that haven't landed yet, like the memory and performance widgets. Also, lately I find myself diving into SpiderMonkey and Gecko work for our upcoming memory and performance tools, and my lack of experience there slows me down considerably. Nevertheless, this is something that bothers me and I intend to improve upon this year.

The second observation [2] is that I'm doing more reviews as time passes, which is to be expected I guess. Last year I did 240 reviews, compared to 220 reviews the year before and 25 reviews in my first year. Obviously those who are around longer get to do more reviews than those who are new to the project, but what I mostly find interesting about this is that I get to do approximately one review per working day. Since I'm counting these from commit messages, they don't take into account the back-and-forth between the patch author and reviewer while shaping up the patch. I counted 372 review requests in my email during the past year (not including feedback or needinfo requests), so I seem to be giving reviews or feedback more like twice a day, every day. This must have contributed in part to the diminished number of patches that I landed myself, but on the flip side I get more visibility in incoming changes and can be more up to speed on my own work, which should decrease the turnaround for reviews on my patches. In addition to that, when I'm doing a review I'm looking at the patch from many different angles (from code correctness and code style issues, to API clarity and architectural direction) and that gives me insights into how our architecture should evolve, which in turn results in more followup work (incidentally, I've filed precisely 100 bugs last year).

Another very important realization is that even if I had spent exactly zero time reviewing other people's work, it would have been very unlikely that I could have managed to land all those 303 patches myself (240 plus 63 of my own). Which is another way to say that we all make a bigger impact by enabling others to do better work, than by strictly fine-tuning our own work output. This is particularly important for achieving a high bus factor for the project, but it is also a necessary ingredient if we want to nurture a community of contributors that will keep the project successful in the years to come. I find this sufficiently important and motivating that I don't plan to be doing less reviews, just try to invest my time more efficiently. We'll see how that goes.

Speaking of nurturing new contributors, I saw in a relevant dashboard that I mentored 8 bugs from 6 new contributors. I'm not sure about the time period of that data, but I've only heard about this dashboard recently, so I assume it should be relevant to this post. Clearly I need to do better in that department.

This past year I've also invested more time in giving public talks and helping with the Firefox OS launch in Greece. The social aspect was extremely rewarding and I've met a ton of great people, so I'd like to do more of that. Not sure how I could combine this with increased code productivity though.

All in all I feel it was a good year and I can't wait to see what the future holds. Onwards!

[1]: I used this command to count my patches (omitting the wc invocation allowed me to inspect the commit messages):
hg log -u past@mozilla.com -d "2013-04-26 to 2014-04-25" --template '{date|isodatesec} : {desc|strip|firstline}\n'|wc -l

[2]:I used the following command to count my reviews. Clearly searching for the right strings is a work of art:
hg log -d "2013-04-26 to 2014-04-25" -k ",past" -k "r=past" -k "r=panos" -k ",panos" -k "r=pastithas" -k ",pastithas" --template '{date|isodatesec} : {desc|strip|firstline}\n'|wc -l

JavaScript debugging updates

2012-11-06T17:39:00.001+02:00

There have been a lot of interesting developments in the JavaScript debugging front since my last update, so let me give you a brief overview of the most important changes.

Let's begin with the eagerly awaited Firefox OS. Starting the debugger in Firefox OS has gotten way easier. There is a toggle in the settings app that will set or reset the devtools.debugger.remote-enabled pref, so that you don't have to mess with files and adb at all. At least for production builds that have marionette disabled by default. Developer builds will have to keep disabling marionette for a bit longer. You can find the Remote Debugging toggle in Settings -> Device Information -> More Information -> Developer. Note that you will have to reboot the device (or restart the b2g process) for the change to have effect. Also, as you can see in the screenshot, it's a good idea to disable out-of-process support while debugging, at least until the debugger can support it.

Firefox OS settings

While we are on the subject of debugging JavaScript on mobile, I shouldn't neglect to mention that Mihai Sucan has landed web console support for Firefox OS and Firefox for Android. There are still a few rough edges, particularly in the case of Firefox OS, but now you can connect your desktop Firefox to your mobile device and see the errors and logged messages, as well as evaluate JS code in the context of the mobile page.

Speaking of desktop Firefox, the amazing Victor Porof has landed an obscene number of patches in the last few weeks, with many fixes and tweaks for the debugger frontend. Of those I'd like to highlight the new functionality to search for variables in the current scope chain while the debugger is paused. Prepend your search with a star to search in the variables view. When you deal with large objects with long scope chains, finding the variable you are looking for can be like searching for a needle in a haystack. Now it's a piece of cake.

Searching for variables

And last, but certainly not least, chrome debugging support has appeared in Nightly this past weekend! This will make debugging add-ons or Firefox itself way easier than ever before. You will need to enable remote debuging (devtools.debugger.remote-enabled) and browser chrome debugging as well (devtools.chrome.enabled, the same pref you switch to get Scratchpad to run against chrome).

Chrome debugging preferences

This will get you a nice new Browser Debugger menu item in the Web Developer menu, which launches a separate debugger process to inspect your browser instance.

Browser Debugger menu item

There has never been a better time to dive into Firefox internals and become a valued contributor. Take a look at this bug, to see how simple debugging has just become. Help us make the web better, by improving the lives of more than 400 million users!

Debugging Firefox OS

2012-10-05T16:18:00.002+02:00

A few months ago, long before the Firefox debugger had shipped, I had tried to see what it would take to get it to debug B2G, or as we call it now, Firefox OS. The hack proved successful and with time it grew into an important debugging target for us. I demoed it at JSConf last April, but at that time a lot of the debugger frontend functionality had not been available in nightlies. As of yesterday, however, nightly builds of Firefox can debug nightly builds of Firefox OS, both in the device and desktop versions. Let me give you a quick rundown of the necessary steps.

First of all you need to go to about:config on your desktop Firefox and turn devtools.debugger.remote-enabled to true. After restarting Firefox you will have a new Remote Debugger menu item in your Web Developer menu.

You need to change the same setting in Firefox OS as well, and that requires finding the prefs.js file. For desktop builds, this will be in gaia-repository/profile/prefs.js. For devices, you should connect it to the desktop through a USB cable, use adb shell to connect to the device, and find the name of the default profile in /data/b2g/mozilla/. The prefs.js file wil be in that subdirectory, which in my case is /data/b2g/mozilla/ngsweij3.default/prefs.js. You can get that file off the device using:

adb pull /data/b2g/mozilla/ngsweij3.default/prefs.js prefs.js

That will store a copy of that file in your current working directory. You need to modify this file and then send it back to the device with:

adb push prefs.js /data/b2g/mozilla/ngsweij3.default/prefs.js

The necessary changes to that file are the following two lines:

user_pref("devtools.debugger.remote-enabled", true);
user_pref("marionette.defaultPrefs.enabled", false);

They enable the debugger server and at the same time disable the Marionette server, to avoid a conflict that we haven’t resolved yet.

That is all that is necessary for the desktop Firefox OS case, but you need to do one more thing for debugging Firefox OS on the device. If you are going to debug over a USB connection (which will also provide helpful console logs via adb logcat) you will need to forward a local port to the device using adb:

adb forward tcp:6000 tcp:6000

Port number 6000 is the default and if you need to change it for some reason, you can do so with the following pref:

user_pref("devtools.debugger.remote-port", 7000);

Debugging over WiFi is also possible, although disabled by default, so you will need to change the following pref to do so:

user_pref("devtools.debugger.force-local", false);

In that case, you don’t have to do the port forwarding step, but you may change the port number in the same way.

In order for the new settings to take effect in the device you have to restart it with ‘adb reboot’.

That’s all folks! Now you just start your nightly Firefox and select the Remote Debugger menu item. A dialog will pop up asking for the remote address and port number to connect to. The defaults will work for the desktop Firefox OS or the device debugging-over-USB cases (unless you changed the port number of course!). If you opted for debugging over WiFi, you will need to provide the IP address of the device. You can find it by tapping on the connection name in the settings.

After the connection is established, you will see a list of the sources loaded on the device, giving you the option to inspect the code, set breakpoints, step through the execution or even modify the values of variables. Here is a screencast that shows the debugger in action:

A better introductory guide to debugging Firefox OS will be available soon on MDN. This is just the first step in supporting Firefox OS development with Firefox developer tools. Just around the corner are web console support and profiling, with more to follow in the near future. If you have issues or suggestions for these tools, please get in touch! Post a comment, file a bug or pop in #devtools for a chat. Let's make the web the best platform to develop on!

Debugging Firefox

2012-10-01T22:34:00.000+02:00

Last week we had a Firefox Developer Tools team get-together at the Mozilla London office. These meetups are an extremely valuable tool for the project to tackle hard problems, because when you put all of the experts on a particular hard problem in the same room, hard problems tend to run away screaming.

We managed to attack a lot of hard problems this week: a profiler, source maps, the developer toolbox, web console remoting and chrome debugging, were just a few of them. And not only that, but we also landed a few exciting new features as well, with fullscreen scratchpad and markup panel preview being my personal favorites. I'm sure others will cover the rest in appropriate detail, but I'd like to say a few things about chrome debugging.

Last March, during the previous team meetup we had managed to get a chrome debugging proof-of-concept working. We were thrilled by it, even though it was a big hack, because the implications were enormous. Something that not many people know is that the Firefox frontend is built in JavaScript, CSS and XUL, which is an HTML derivative with additional capabilities. That is, we use the exact same technologies that power your favorite web applications, to build the browser that displays said web apps to the user. This similarity has always been front and center to our thinking when considering tools for web developers: how can we use these tools to make hacking on Firefox easier?

Without easy to use tools, fixing bugs is time-consuming and adding features demands a very disciplined approach to programming. It hampers our ability to move the project faster and innovate rapidly. Furthermore, having a non-profit foundation to back the development of an open-source browser means that you don't have the option to simply throw more money at the problem, hiring more engineers than everybody else to increase the pace of innovation in the product. This is a browser created by thousands of contributors to take care of their own interests. We have to make their work easier, by giving them the tools they need to work faster.

When we did the chrome debugging prototype last March, the Firefox Debugger had been in trunk for a short while, disabled by default. No visual styling had been done on it and only the most basic of functionality was there. Six months later we have shipped the first version and we keep iterating on it.

The proof-of-concept couldn't debug chrome window globals, only modules, and was hijacking the remote debugging protocol in a hackish way that was guaranteed never to make it into a Firefox release. This time we have both kinds of globals supported and a sound protocol design that was agreed upon a few months ago.

This new tool, which we dubbed "Browser Debugger", will be useful to add-on developers, too. In my demo last week I used it to debug Firebug. We have patches under review or close enough that implement a first draft of the experience we aim to present users with. I want to land that as soon as we can, enable the feature by default (but conditioned on the devtools.chrome.enabled pref like other tools) and keep iterating.

It was without a doubt a very rewarding week. Some might expect us to also treat these rare events as opportunities to hang out, get drunk and take embarrassing pictures of each other. I am not going to either confirm or deny such allegations. At least not until I'm presented with hard, photographic evidence. Preferably on Flickr.

JSConf 2012

2012-04-06T20:16:00.000+02:00

This week I had the opportunity to attend JSConf. I've been trying to make it to one of these conferences in the past without success and I now realize I was missing a lot! Chris and Laura Williams deserve special credit for the excellent job they are doing here. Even though I missed all the parties (because jetlag), the venue, the speakers, the tracks, and the conversations were all top class. Highly recommended.

If I had to pick the highlights of the conference, an undoubtedly difficult undertaking due to the sheer number of awesome talks, my personal favorites would be these two:

The B2G phones given out by Mozilla to all attendees, a gesture that sparked the creativity of quite a few people, culminating in a few B2G apps after a few hours of hacking. Yes, it's that easy.

The chance to meet and listen to Dan Ingalls, in my humble opinion one of the greatest programmers of our time, giving his inspirational talk on the Lively Kernel. It would be a failure of our profession if the ideas in Lively Kernel do not find a way to reach mass market at some point. This system, like its predecessors Squeak and Dynabook, represent a more humane way of programming computers that could be a game changer.

My own Debugging B2G talk was well-attended, even though a malfunctioning video cable threatened to kill it before it even began. I received lots of comments afterwards and requests for access to the debugger build that I demoed, and even though I pointed people to the patches, I think that the best thing to do for those interested is to wait a few more weeks, until we get all of that work in nightlies. After getting back at the hotel I recorded a screencast of the talk for those who couldn't attend, but unfortunately you will miss me mumbling while frantically plugging and unplugging a video cable and cursing at thunderbolt ports. Hands down the highlight of the talk.

<p>&amp;amp;amp;amp;amp;lt;p&amp;amp;amp;amp;amp;gt;&amp;amp;amp;amp;amp;amp;amp;amp;amp;lt;p&amp;amp;amp;amp;amp;amp;amp;amp;amp;gt;&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;lt;a target='_blank' href='http://vid.ly/7o0x1p'&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;gt;&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;lt;img src='http://cf.cdn.vid.ly/7o0x1p/poster.jpg' /&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;gt;&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;lt;/a&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;gt;&amp;amp;amp;amp;amp;amp;amp;amp;amp;lt;/p&amp;amp;amp;amp;amp;amp;amp;amp;amp;gt;&amp;amp;amp;amp;amp;lt;/p&amp;amp;amp;amp;amp;gt;</p>

All in all it was a great opportunity to meat people, have conversations on the present and future of Firefox developer tools and ride a bull. Too bad I missed that last one.

Debugging mobile phones

2012-02-19T19:30:00.003+02:00

Last week I posted an update on the progress we are making towards a functional JavaScript debugger in Firefox. We still have a ways to go before we get there, but the foundation this is based on, namely the Remote Debugging Protocol, provides a solid basis for debugging mobile browsers, and as it turns out, even mobile operating systems. Supporting these use cases is still farther down our list, but I decided to spend a few hours the other day trying to estimate how far we are from this goal. It turns out we are not that far, and I have a screencast to prove it.

I had done some prototyping work for Firefox for Android a while back, which was rather straightforward, so I was curious to know what would it take to tackle a more ambitious project: debugging Boot to Gecko. After a few hours of hacking and orienting myself around the B2G and Gaia code base, I got this far:

Not too bad I think for a few hours work. Don't expect this to be ready of course, before we have a working debugger in desktop Firefox. If, however, you'd like to follow along and maybe even lend a hand, you are always welcome!

Debugging JavaScript

2012-02-13T10:08:00.000+02:00

tl;dr: Firefox nightlies now ship with an experimental JavaScript debugger. It's not ready for end-users yet, but we are feverishly working on filling in the missing bits.

Debugging is always hard, especially when debugging other people's code. Adding a bunch of console.log() statements, becomes tedious after a while and that's assuming that you are familiar enough with the code in question to know where to place them. Tools like Firebug are a tremendous help in this case. Having the ability to pause the execution at any point or inspect variables and stack frames, provides valuable insights into the runtime behavior of a program.

Firebug depends on the traditional JSD API in SpiderMonkey to perform its magic, but nowadays JSD is limiting in a number of ways. In order to overcome these limitations, the new JSDBG2 API was designed and implemented by Jim Blandy and Jason Orendorff, and the first version of this work landed in Firefox 8. Jim, being the kind of guy that has forgotten more about debuggers than I will ever know, designed a remote debugging protocol to separate the debugger from the browser, allowing for debugging desktop browsers, mobile browsers, mail readers and even more exotic software in the future. Dave Camp got a great protocol implementation up and running, but was assigned managerial duties (poor sap!) before having a chance to finish the job, so Victor Porof, Mihai Șucan and yours truly lined up to carry the torch. The culmination of this work landed in mozilla-central last week, including the remote protocol implementation and a prototype UI (disabled by default). It is now available in nightlies for adventurous souls to play with.

Let me be clear about that last point: what's in there right now is not useful for day-to-day work, even for experienced developers. There is no UI to add breakpoints besides the experimental Graphical Command Line Interface (whose awesomeness deserves a separate post), no stepping, no variable inspection besides call parameters and 'this', and numerous other limitations. In short: it's not ready, yet. In the course of the next weeks we will be furiously working to add the missing bits and improve the UI, so that it becomes a worthy addition to our existing suite of developer tools.

If, however, you are the kind of person who doesn't take no for an answer, here is what you need to do to play with it: type about:config in the awesomebar and set the preference devtools.debugger.enabled to true. Also, set devtools.gcli.enable to true in order to be able to set breakpoints. Restart your browser and select Tools -> Web Developer -> Script Debugger to open it. You will need to reload the page for the debugger to become aware of the scripts in the page (see? It's not ready!). Browse through the source scripts to find a place where you would like to add a breakpoint, then open the web console and type break add line <url> <line>. Don't worry about memorizing it, GCLI will guide you along the way, suggesting among the available options at each step. When the code hits the breakpoint, the debugger will display the stack frames and the variables in scope (not all of them, it's not ready, remember?) for inspection.

If you have better things to do with your time than play with experimental, uninished debuggers, but would still like to take a peek into what's in store already, here is a short screencast that should give you a broad idea. Note that the stepping functionality in the demo is not in the nightlies yet, because, you guessed it, it's not quite ready.

Ready or not, we would love to have any feedback you might have. Drop by #devtools for a chat, or file bugs to ~~torture~~help us get things fixed. Rest assured, when it's done, debugging web apps will never be the same again.

Next stop: Mozilla

2011-04-16T21:00:00.002+02:00

Six months ago I left my previous job, in order to join a product company and experience new things. The move was rewarding in many ways, but most importantly, because of the people I've met. We've accomplished a lot together and laid out the foundations for much more. As I've established before, this is not the right time to be moving on, but while in theory, theory and practice are the same, in practice, they are not. When good fortune knocks on your door, it's a fool who does not answer.

So I am seizing the opportunity to work with some of the best hackers in the world, on a product that I adore, with technologies that until now I had to stay up late at night to use. In a few days I’ll be joining Mozilla to work on Firefox Developer Tools, a passion of mine for quite some time.

I am thrilled with this opportunity to learn from the best, and have an impact on the lives of millions. And to do this with a non-profit organization, whose mission is to promote openness, innovation and opportunity on the web, for the public good.

I'm almost wishing I gave-up this long-needed vacation.

Moving On

2010-10-31T01:46:00.002+02:00

About 10 years ago, I was invited to become employee No1 of a new software startup. I didn’t know much about startups back then, besides that they were supposedly cool, and employees weren’t treated like small cogs in a giant machine. Indeed, it felt really fun to build our first product in my 5 year-old laptop, running FreeBSD, of all things. That, and getting praised by our users, too.

Things got even better as time went by: interesting projects to build, exciting technologies to explore, new colleagues to work with, new challenges to face. I felt my work was making a real impact, getting the team to follow new and exciting technological paths, often far away from the mainstream. I became partner, pushed for launching our first large consumer product, coded, mentored, planned and dreamed about where to go next. I had a 27-inch Mac, a big desk and a great view from my office window.

And then I left.

When is it a good time to leave? When everything is over, it’s obviously too late. When things are just starting to take shape, it’s definitely too soon. I used to say, ‘when you are done’. But it turns out that, in real life, you are never done. There are always more things to do, unfinished businesses to attend to, and existing plans to follow through. Perhaps a better answer is: ‘when you are too comfortable, almost on the brink of getting bored’.

A friend asked me why would I leave this company in what appears, by all accounts, to be the pinnacle of our accomplishments, ‘now that your work is being recognised and respected’, as he put it. Another said that we’re all going through some pretty hard times, and it wouldn’t be wise to take risks. A rather unsettled family member asked: ‘aren’t you afraid?’

Of course I was.

Afraid of leaving behind the tried-and-true for something new and unknown; afraid that the new gig might not live up to the initial impressions; afraid that my work-life balance might deteriorate. However, I’ve learned long ago that fear isn’t something to be afraid of. As Eleanor Roosevelt said once: ‘You gain strength, courage, and confidence by every experience in which you really stop to look fear in the face. You must do the thing which you think you cannot do.’ What I wanted to do, is to become even better at my craft. To have an impact on even more people’s lives through my work. To inspire and be inspired. To be afraid and excited again.

So, as of a few days ago I’ve taken up an offer to work for Golden Deal. It’s a product company, something I’ve always been keen on, with a great track record and an awesome crew that is rapidly expanding (are you a senior dev or admin? we’re hiring!). I’m still getting to know the place around here, but it feels good, and I think we have a shot at becoming one of the greatest Internet companies in Greece.

I know that my former teammates will continue to rock on, and I’ll be cheering from the sidelines for their future accomplishments. As for me, I just might find more time to blog again.

P.S.: For folks contemplating a career move, here are some tips for choosing your next gig. This is not how I approached my own case, since I wasn’t really on the market for a job. I was more like seduced, from some pretty persuasive folks. However, had I been looking for a new employer, this is what I would have looked out for:

Product company: this is where you have a better chance at learning the ropes of modern technologies, without doing pure academic exercises that have little to no importance for the business.
Startup mentality: you want to be a part of a group that fights as a whole for the success of the company, not a bunch of bureaucrats isolated in their silos.
Experienced team with a good track record: so you can learn from people better, smarter or wiser than you, and improve your skills.
Teams not afraid to experiment with new and unproven technologies: because innovation is the name of the game, at least for startups and product companies.
A ‘getting things done’ attitude: because success apparently only comes for the right kind of people.
Honest presentation: this is rather obvious.

Bear in mind that people have a tendency to present a rosy picture of their situation, because, if nothing else, it makes them feel better about themselves. Be a little pessimist and assume things are a tad worse, and you’ll be closer to the truth most of the time. Also, do some research on what you’re being told, don’t trust their sales pitch blindly.

I know I did.

GSS/Pithos update: iPhone, Android, Mongo & getting to 2.0

2010-10-12T22:55:00.000+02:00

It has been a while since my last post on GSS/Pithos, and there have been quite a few important announcements that I have not blogged about, only tweeted:

a shiny native iPhone client
a brand new Android client
a back-office application for statistics and admin tasks
public folders for hosting static web sites
a slew of performance, usability and bug fixes for the web & desktop clients
new user registration and quota upgrade workflows

And since the code base has been adopted by other European NRNs, we have also improved our release engineering process and documentation. This however was just the tip of the iceberg. What we are now close to completing, is nothing short of an almost complete rewrite of the core GSS server, aiming at even greater performance and scalability. We are about to release GSS 2.0.

The initial plan was far more modest. We had identified the PostgreSQL installation that stores all of the the system’s data besides the actual files, as a future bottleneck when user adoption would start to accelerate. In such an event there were a few low-to-medium-cost solutions that we could pursue, like replication and manual sharding, but since the problem seemed best suited for a NoSQL solution, we decided to bite the bullet and attempt a transition.

Our field search included lots of products, like Mongo, Couch, Cassandra, Riak, Redis, HBase, MemcacheDB and others. The basic requirements were high write throughput, sharding, replication and easy migration. Most of the products boasted high performance for writes and lots supported replication. In the end we discarded the ones that didn’t support sharding, which limited the options a lot, and we investigated the various APIs, trying to figure out what it would mean to rewrite GSS for one of these. In the end we chose Mongo for its rich query support, since it was the best fit for our problem at hand, making the transition of the existing code base easier. Although we’d love to have had the opportunity to try different ports for each of those datastores, it is most likely that we would have had to rewrite the entire server from scratch, which was not an option.

The initial plan was to try to isolate the changes to the data access layer, essentially building a new adapter for the middle tier that would appear almost identical to the existing code. This effort was fruitless however, since we could not find satisfactory solutions for various issues, like transactions, data mapping, entity lifecycle, etc. We came to the conclusion that even if we managed to build a not-too-leaky abstraction for Mongo, it would have probably killed the performance, making the switch from PostgreSQL pointless.

So we ditched JPA and container-managed transactions and decided to make a simpler POJO mapping to Mongo using Morphia, a Java data mapper for Mongo that takes care of the nitty-gritty details and repetitive coding, for only a minor performance hit. Since Mongo only supports atomicity for a single query, we reverted to manual transaction handling and decided to ditch the EJBs that formed the backbone of GSS altogether, since they didn’t buy us anything at that point. The new GSS server would be a set of servlets, service and domain objects, wired together via Google Guice. Since we could host the new server in a plain servlet container, we went from JBoss to Jetty, which vastly reduced the server startup time from 60 seconds down to 2. As a consequence, development round-trip times went down, and coding was fast again. Lots of object copying and conversions from entities to DTOs and back again were now unnecessary, since there was no more a transaction boundary at the session bean layer. This kept memory usage low, increased cache locality and boosted performance. There are lots of interesting details about our approach, but they would be better discussed in separate blog posts. If you are interested, you can already check out the code in the gss2 branch of the GSS repository.

The next generation GSS server is not production ready yet, but we'll be getting there soon, and so far it has been a great ride!

The Cloud Desktop

2010-02-19T22:23:00.006+02:00

This was originally posted in the mynetworkfolders.com blog.

We live in a browser these days. E-mail, word-processing, spreadsheets, chat, are all occupying a browser window in front of me. Maybe not for everyone yet, but the trend is clear, and has been clear for quite some time now. This is a good thing. The Web of today bears little similarity to the Web of the millennium. It's fast, it's powerful and it's ubiquitous. A consequence of that evolution is the simplicity with which I can continue my work when leaving my Linux workstation at work and sit in front of my Mac at home. Since my applications are web-based, they are OS- and browser-agnostic. More importantly, my data are always there when I need them. Readily accessible from my desktop, laptop or mobile phone. I may keep local backups for archival purposes, but my master copies are always online.

Assuming this trend is indeed inescapable, one can't help but wonder: what are the applications that have not migrated to the cloud, yet? What kind of things would I like to do online, seamlessly, from every device I own? What are the barriers that make our current online experience less powerful than the desktop one?

One thing that we noticed was missing from the online desktop, was the... desktop! Yes, my spreadsheets are online. Sure, my pictures are online. Yes, I can upload arbitrary files in various storage services. But, I couldn't find a way to use a familiar file manager interface to work with my data in the cloud, as I do on my computer. This was the need we decided to fulfill with MyNetworkFolders.

MyNetworkFolders attempts to provide an answer to the question: how would my desktop look in the cloud? And even though we are nowhere near done yet, the results so far are exciting and very promising.

MyNetworkFolders offers a familiar file manager interface for my cloud storage needs, with advanced functionality and ease of use. Documents can be viewed, modified, searched or sent to the trash. They can be shared with other users or among the members of a group using flexible ACLs, or even be publicly accessible from the whole internet. They can be versioned for tracking changes more effectively. The service can be accessed from a variety of ways, including a web browser, a desktop client, a WebDAV network share or (really soon) a mobile phone. But most importantly, there is no vendor lock-in. There is a REST-like API available, battle-tested from all the above client implementations, giving everyone the opportunity to access his data in his own particular way. Or even get them out of our service altogether.

Oh, and one other thing. It's all based on open-source software. The core infrastructure, the various clients we have built, even API helpers in various programming languages, are all available for anyone and everyone to see, use and modify. Or maybe used to setup and run YourNetworkFolders, if you feel really ambitious.

If however you just want a true desktop environment, with unparalleled flexibility, and maybe also see how far the cloud desktop rabbit-hole goes, then come along. It's going to be an exciting ride.

Lessons from Startup Weekend

2009-12-07T23:50:00.002+02:00

I had an exhausting but fun weekend at the Athens Startup Weekend a few days ago. Along with Christos I joined Yannis, Panagiotis Christakos and Babis Makrinikolas on the Newspeek project. When Yannis pitched the idea on Friday night, the main concept was to create a mobile phone application that would provide a better way to view news on the go. I don't believe it was very clear in his mind then, what would constitute a "better" experience, but after some chatting about it we all defined a few key aspects, which we refined later with lots of useful feedback and help from George. Surprisingly, for me at least, in only two days we managed to design, build and present a working prototype in front of the judges and the other teams. And even though the demo wasn't exactly on par with our accomplishments, I'm still amazed at what can be created in such a short time frame.

Newspeek, our product, had a server-side component that periodically collected news items from various news feeds, stored them and provided them to clients through a simple REST API. It also had an iPhone client that fetched the news items and presented them to the user in a way that respected the UI requirements and established UX norms for that device.

So, in the interest of informing future participants about what works and what doesn't work in Startup Weekend, here are the lessons I learned:

If you plan to win, work on the business aspect, not on the technology. Personally, I didn't go to ASW with plans to create a startup, so I didn't care that much about winning. I mostly considered the event as a hackathon, and tried my best to end up with a working prototype. Other teams focused more on the business side of things, which is understandable, given the prize. Investors fund teams that have a good chance to return a profit, not the ones with cool technology and (mostly) working demos. Still, the small number of actual working prototypes was a disappointment for me. Even though the developers were the majority in the event, you obviously can't have too many of them in a Startup Weekend.
For quick server-side prototyping and hosting, Google App Engine is your friend. Since everyone in the team had Java experience, we could have gone with a JavaEE solution and set up a dedicated server to host the site. But, since I've always wanted to try App Engine for Java and the service architecture mapped nicely to it, we tried a short experiment to see if it could fly. We built a stub service in just a few minutes, so we decided it was well worth it. Building our RESTful service was really fast, scalability was never a concern and the deployment solution was a godsend, since the hosting service provided for free by the event sponsors was evidently overloaded. We're definitely going to use it again for other projects.
jQTouch rocks! Since our main deliverable would be an iPhone application, and there were only two of us who had ever built an iPhone application (of the Hello World variety), we knew we had a problem. Fortunately, I had followed the jQTouch development from a reasonable distance and had witnessed the good things people had to say, so I pitched the idea of a web application to the team and it was well received. iPhone applications built with web technologies and jQTouch can be almost indistinguishable from native ones. We all had some experience in building web applications, so the prospect of having a working prototype in only two days seemed within the realm of possibility again. The future option of packaging the application with PhoneGap and selling it in the App Store was also a bonus point for our modest business plan.
For ad-hoc collaboration, Mercurial wins. Without time to set up central repositories, a DVCS was the obvious choice, and Mercurial has both bundles and a standalone server that make collaborative coding a breeze. If we had zeroconf/bonjour set up in all of our laptops, we would have used the zeroconf extension for dead easy machine lookup, but even without it things worked flawlessly.
You can write code with a netbook. Since I haven't owned a laptop for the last three years, my only portable computer is an Asus EEE PC 901 running Linux. Its original purpose was to allow me to browse the web from the comfort of my couch. Lately however, I'm finding myself using it to write software more than anything else. During the Startup Weekend it had constantly open Eclipse (for server-side code), Firefox (for JavaScript debugging), Chrome (for webkit rendering), gedit (for client-side code) and a terminal, without breaking a sweat.
When demoing an iPhone application, whatever you do, don't sweat. Half-way through our presentation, tapping the buttons didn't work reliably all the time, so anxiety ensued. Since we couldn't make a proper presentation due to a missing cable, we opted for a live demo, wherein Yannis held the mic and made the presentation, and I posed as the bimbo that holds the product and clicks around. After a while we ended up both touching the screen, trying to make the bloody buttons click, which ensured the opposite effect. In retrospect, using a cloth occasionally would have made for a smoother demo, plus we could have slipped a joke in there, to keep the spirit up.

All in all it was an awesome experience, where we learned how far we can stretch ourselves, made new friends and caught up with old ones. Next year, I'll make sure I have a napkin, too.

Spell-checking in JavaScript

2009-08-20T14:06:00.002+03:00

I just heard of Atwood's Law a few days ago. Apparently Jeff Atwood first published his discovery a couple of years ago, which is like a century in Internet time, but I can't keep up with everything. The Law states that "any application that can be written in JavaScript, will eventually be written in JavaScript". As Atwood explains, it is based on Tim Berners-Lee's Principle Of Least Power:

Computer Science spent the last forty years making languages which were as powerful as possible. Nowadays we have to appreciate the reasons for picking not the most powerful solution but the least powerful. The less powerful the language, the more you can do with the data stored in that language.

I think there is a common theme between Atwood's Law, Zawinski's Law ("Every program attempts to expand until it can read mail. Those programs which cannot so expand are replaced by ones which can") and Greenspun's Tenth Rule ("Any sufficiently complicated C or Fortran program contains an ad hoc, informally-specified, bug-ridden, slow implementation of half of Common Lisp"). The inevitability of the specified outcome makes the world seem like a much simpler place. There is no need to argue. Believe and ye shall be redeemed.

I might have been in this state of mind, when I came across Peter Norvig's "How to Write a Spelling Corrector", because I knew instantly I had to port the algorithm to JavaScript. The algorithm is quite simple and has already been ported to many different languages, so I seized the opportunity to study the differences in expressiveness, performance and style, between these languages and my latest affection, JavaScript.

After a few night's work I have it working and free (as in beer and speech) for anyone to use. Check out the speller project on GitHub. If you want to understand how the algorithm works, you should go and read Norvig's article, although you might get some hints from the comments in my code. The modestly condensed version of the code is 53 lines:


var speller = {};
speller.train = function (text) {
  var m;
  while ((m = /[a-z]+/g.exec(text.toLowerCase()))) {
    speller.nWords[m[0]] = speller.nWords.hasOwnProperty(m[0]) ? speller.nWords[m[0]] + 1 : 1;
  }
};
speller.correct = function (word) {
  if (speller.nWords.hasOwnProperty(word)) return word;
  var candidates = {}, list = speller.edits(word);
  list.forEach(function (edit) {
    if (speller.nWords.hasOwnProperty(edit)) candidates[speller.nWords[edit]] = edit;
  });
  if (speller.countKeys(candidates) > 0) return candidates[speller.max(candidates)];
  list.forEach(function (edit) {
    speller.edits(edit).forEach(function (w) {
      if (speller.nWords.hasOwnProperty(w)) candidates[speller.nWords[w]] = w;
    });
  });
  return speller.countKeys(candidates) > 0 ? candidates[speller.max(candidates)] : word;
};
speller.nWords = {};
speller.countKeys = function (object) {
  var attr, count = 0;
  for (attr in object)
    if (object.hasOwnProperty(attr))
      count++;
  return count;
};
speller.max = function (candidates) {
  var candidate, arr = [];
  for (candidate in candidates)
    if (candidates.hasOwnProperty(candidate))
      arr.push(candidate);
  return Math.max.apply(null, arr);
};
speller.letters = "abcdefghijklmnopqrstuvwxyz".split("");
speller.edits = function (word) {
  var i, results = [];
  for (i=0; i < word.length; i++)
    results.push(word.slice(0, i) + word.slice(i+1));
  for (i=0; i < word.length-1; i++)
    results.push(word.slice(0, i) + word.slice(i+1, i+2) + word.slice(i, i+1) + word.slice(i+2));
  for (i=0; i < word.length; i++)
    speller.letters.forEach(function (l) {
      results.push(word.slice(0, i) + l + word.slice(i+1));
    });
  for (i=0; i <= word.length; i++)
    speller.letters.forEach(function (l) {
      results.push(word.slice(0, i) + l + word.slice(i));
    });
  return results;
};

It may not be as succinct as Norvig's Python version (21 lines), or the record-holding Awk and F# versions (15 lines), but is much better than C (184 lines), C++, Perl, PHP, Rebol and Erlang. There is even a Java version with 372 lines. It must contain some sort of spell-checking framework in there, or something. The condensed version above, although correct, has terrible performance in most JavaScript engines, however. For real-world use you should pick the regular version which may be slightly longer, but performs much better.

Since this was a toy project of mine, I wanted to play with ServerJS modules as well, in order to run it as a shell script. I turned the code into a securable module, so you can run it from the command line, using narwhal. I have a couple of scripts to that end. Of course since this is JavaScript, you can try it from your browser, by visiting the demo page, courtesy of GitHub Pages. Be sure to use a modern browser, like Firefox 3.5, Safari 4 or Chrome 3 (beta), otherwise you won't be able to run the test suite, since I used the brand-new Web Workers to make the long-running tasks run in the background.

Norvig's Python implementation took 16 seconds for test 1 and the best I got was 25 seconds with Safari 4 on my Mac. Narwhal is using Rhino by default, so it is definitely not competitive in such tests (139 seconds), but I'm planning to fix support for v8cgi and give that a try.

And it goes without saying that I'd love to hear about ways to improve the performance or the conciseness of the code. If you have any ideas, don't be shy, leave a comment or even better fork the code and send me a pull request on GitHub.

GSS authentication

2009-07-07T16:23:00.001+03:00

As I've described before, the GSS API is a REST-like interface for interacting with a GSS-based service, like Pithos. Using regular HTTP commands a client can upload, download, browse and modify files and folders stored in the GSS server. These interactions have two important properties: they store no client state to the server and they are not encrypted. I have already mentioned the most important benefits from the stateless architecture of the GSS protocol, namely scalability and loose coupling along the client-server boundary. In the same line of thought, SSL/TLS encryption of the transport was avoided for scalability reasons. Consequently, these two properties of the communication protocol, lead to another requirement: message authentication for each API call.

Since no session state is stored in the server, the client must authenticate each request as if it were the first one. Traditional web applications use an initial authentication interaction between client and server, that creates a unique session ID, which is transmitted with each subsequent request in that same session. While using this ID, the client does not need to authenticate again to the server. Stateless protocols, like WebDAV for instance, cannot rely on such an ID and have to transmit authentication data in each call. Ultimately the tradeoff is a minor increase in the message payload, in return for a big boost in scalability. The HTTP specification defines an Authorization header for use in such cases and WebDAV uses the HTTP Digest Access Authentication scheme. The GSS API uses a slight variation in that theme, blended with some ideas from OAuth request signing.

Essentially the standard HTTP Authorization header is populated with a concatenation of the username (for identifying the user making the request) and a HMAC-SHA1 signature of the request. The request signature is obtained by applying a secret authentication token to a concatenated string of the HTTP method name, the date and time of the request and the actual requested path. The GSS API page has all the details. The inclusion of the date and time helps thwart replay attacks using a previously sniffed signature. Furthermore, a separate header with timestamp information, X-GSS-Date, is used to thwart replay attacks with a full copy of the payload. The authentication token is issued securely by the server to the client and is the time-limited secret that is shared between the client and server. Since it is not the user password that is used as a shared secret, were the authentication token to be compromised, any ill effects would have been limited to the time period the token was valid. There are two ways for client applications to obtain the user's authentication token, and they are both described in detail in the API documentation.

In my experience, protocol descriptions are one thing and working code is a totally different one. In that spirit, I'm closing this post with a few tested implementations of the GSS API signature calculation, for client applications in different languages. Here is the method for signing a GSS API request in Java (full code here):


public static String sign(String httpMethod, String timestamp,
        String path, String token) {
  String input = httpMethod + timestamp + path;
  String signed = null;

  try {
    System.err.println("Token:" + token);
    // Get an HMAC-SHA1 key from the authentication token.
    System.err.println("Input: " + input);
    SecretKeySpec signingKey = new SecretKeySpec(
        Base64.decodeBase64(token.getBytes()), "HmacSHA1");

    // Get an HMAC-SHA1 Mac instance and initialize with the signing key.
    Mac hmac = Mac.getInstance("HmacSHA1");
    hmac.init(signingKey);

    // Compute the HMAC on the input data bytes.
    byte[] rawMac = hmac.doFinal(input.getBytes());

    // Do base 64 encoding.
    signed = new String(Base64.encodeBase64(rawMac), "US-ASCII");

  } catch (InvalidKeyException ikex) {
    System.err.println("Fatal key exception: " + ikex.getMessage());
    ikex.printStackTrace();
  } catch (UnsupportedEncodingException ueex) {
    System.err.println("Fatal encoding exception: "
                + ueex.getMessage());
  } catch (NoSuchAlgorithmException nsaex) {
    System.err.println("Fatal algorithm exception: "
                + nsaex.getMessage());
    nsaex.printStackTrace();
  }

  if (signed == null)
    System.exit(-1);
  System.err.println("Signed: " + signed);
  return signed;
}

Here is a method for signing GSS API requests in JavaScript, that uses the SHA-1 JavaScript implementation by Paul Johnston (full code here):


function sign(method, time, resource, token) {
  var q = resource.indexOf('?');
  var res = q == -1? resource: resource.substring(0, q);
  var data = method + time + res;
  // Use strict RFC compliance
  b64pad = "=";
  return b64_hmac_sha1(atob(token), data);
}

Here is a Tcl implementation by Alexios Zavras (full code here):


proc ::rest::_sign {id} {
  variable $id
  upvar 0 $id data
  set str2sign ""
  append str2sign $data(method)
  append str2sign $data(date)
  set idx [string first ? $data(path)]
  if {$idx < 0} {
    set str $data(path)
  } else {
    incr idx -1
    set str [string range $data(path) 0 $idx]
  }
  # append str2sign [::util::url::encode $str]
  append str2sign $str
  puts "SIGN $str2sign"
  set sig [::base64::encode [::sha1::hmac -bin -key $::GG(gsstoken) $str2sign]]
  set data(signature) $sig
}

I'd love to show implementations of the signature calculation for other languages as well, but since my time is scarce these days, I could use some help here. If you've written one yourself and you'd like to share, leave a comment and I'll update this post, with proper attribution of course.

Retrying transactions with exponential backoff

2009-06-17T19:25:00.004+03:00

The method I described in my previous post about retrying transactions could use some improvement. A deficiency that will only become relevant in highly congested servers is the constant retry interval. When two or more transactions try to commit at the same time and fail, with the code from the last post they will retry probably simultaneously again. Random runtime events (like process/thread scheduler decisions, JIT compiler invocations, etc.) might help avoid collisions, but in general the collided transactions may well collide again. And again. And then again, until the specified maximum number of retries is reached.

The general method to alleviate such problems is to randomize the retry intervals. The most well-known algorithm in this category is called exponential backoff. This is one way to implement it for the utility method in my previous post (full code in gss):


public T tryExecute(final Callable<T> command) throws Exception {
  T returnValue = null;
  // Schedule a Future task to call the command after delay milliseconds.
  int delay = 0;
  ScheduledExecutorService executor = Executors.newScheduledThreadPool(1);
  for (int i = 0; i < TRANSACTION_RETRIES; i++) {
    final int retry = i;
    ScheduledFuture<T> future = executor.schedule(new Callable<T>() {

      @Override
      public T call() throws Exception {
        return command.call();
      }
    }, delay, TimeUnit.MILLISECONDS);

    try {
      returnValue = future.get();
      break;
    } catch (ExecutionException e) {
      Throwable cause = e.getCause();
      if (!(cause instanceof EJBTransactionRolledbackException) ||
          retry == TRANSACTION_RETRIES - 1) {
        executor.shutdownNow();
        throw new Exception(cause);
      }
      delay = MIN_TIMEOUT + (int) (MIN_TIMEOUT * Math.random() * (i + 1));
      String origCause = cause.getCause() == null ?
          cause.getClass().getName() :
          cause.getCause().getClass().getName();
      logger.info("Transaction retry #" + (i+1) + " scheduled in " + delay +
          " msec due to " + origCause);
    }

  }
  executor.shutdownNow();
  return returnValue;
}

The notable changes in this version are the delayed invocations of the Callable and the retry interval calculation. The former is accomplished using a ScheduledFuture from the excellent utilities in java.util.concurrent, which gets executed after a random time period. The interval calculation is something that can be implemented in a number of ways, either with monotonically increasing intervals or not. I opted for the formula above since it provided the fastest collision elimination in my tests, much faster than the monotonically increasing interval formulas I tried. The MIN_TIMEOUT constant is more of a black art. It should be tuned to the particular hardware and network setup in order to attain maximum efficiency: the minimum number of retries with the minimum interval between retries.

Another issue raised in a comment in my previous post was that the exception I am using as guard, EJBTransactionRolledbackException, may be too generic for this purpose. This is definitely true, EJBTransactionRolledbackException just wraps an OptimisticLockException as one would expect for this case, which in turn wraps the Hibernate-specific StaleObjectStateException that is thrown initially. However, contention in a database table, index or row might not necessarily result in optimistic locking violations, but in deadlocks as well, when one transaction holds exclusive locks on resources that are needed by the other and vice versa. Deadlock detection is performed by the DBMS, which forces a rollback of both transactions. This time however the initial exception (at least in my testing, might be DBMS or JDBC driver-specific) is GenericJDBCException which gets wrapped in a PersistenceException, which in turn is put inside a EJBTransactionRolledbackException before being received by our call site. Therefore, the generic nature of the guard is not a problem in this case. On the contrary it covers more relevant cases and one might argue even that there is no better service that can be offered to the caller, than retrying in all such occasions.

Retrying transactions in Java

2009-06-15T16:18:00.011+03:00

When building a web-based system, the most often encountered persistence solution is the use of a relational DBMS. This provides many benefits, but probably the most important one is that its behavior and usage are widely understood after so many decades of use. The simplicity of the ACID transactional model, coupled with the nice fit of the HTTP request/response protocol, make for a killer combination.

The simplicity starts to break down however, when the load on the DBMS server increases, causing transactions to start failing. The DBMS can guarantee that no data corruption will occur, but handling such cases is not that simple. The scalable Optimistic Locking model requires transactions to be resubmitted when failure occurs. Since this is supposed to be rather rare (otherwise optimistic locking should probably not be used), developers often cheat by not dealing with such errors, letting them bubble up to the user, who is then responsible for resubmitting the transaction. This leads to simple, elegant codebases and unhappy users.

Adding ORM to the mix only makes things worse. In JavaEE, the JPA API hides much of the complexity of dealing with relational databases, but corner cases become harder. Add the container-managed transaction boundaries provided by session EJBs and one might be tempted to raise his hands up in despair.

However a solution is not as hard as it sounds. Provided that session EJB methods are sufficiently free from side effects, one can replay a transaction for a number of times from the call site (external to the session bean) with a small amount of code and a small added complexity to the call sites. Since I seem to come up against this requirement quite often, I thought I should describe one way to do it. The following tryExecute() helper method, retries any action supplied in a standard Callable interface for a number of times, in the face of optimistic locking violations:


public T tryExecute(Callable<T> command) throws Exception {
  T returnValue = null;
  for (int i = 0; i < TRANSACTION_RETRIES; i++) {
    try {
      returnValue = command.call();
      break;
    } catch(EJBTransactionRolledbackException trbe) {
      if (i == TRANSACTION_RETRIES - 1)
        throw trbe;
      } catch (Exception e) {
        throw e;
      }
      logger.debug("Transaction retry: " + (i+1));
    }
  return returnValue;
}

By abstracting away the session bean method code in a parameterized class we can reuse this method in every place that needs to retry transactions. Using a Callable as a parameter lets us return any values received from the session bean. For supplying bean methods that do not return values, we could create Runnable instances and wrap them to Callable using Executors.callable(). You can see the original code here.

If you use the above code in a JavaSE environment without EJB, you should probably need to catch OptimisticLockException for JPA, or even StaleObjectStateException, if you don't mind depending on a Hibernate API. I use the above helper method like this in my code:


file = new TransactionHelper<FileHeaderDTO>().tryExecute(new Callable<FileHeaderDTO>() {
  @Override
  public FileHeaderDTO call() throws Exception {
    return sessionBean.createFile(user.getId(), folder.getId(), name, mimeType, uploadedFile);
  }
});

Java's verbosity makes this not that easy to understand at first sight, but what happens is very simple indeed: I wrap a one-liner call to the session bean method createFile(), in a Callable and supply it to tryExecute(). The result from the bean method is received in a local variable as usual. One thing to keep in mind though is that since Java doesn't have closures, the above anonymous function can only receive final variables as parameters. In my example user, folder, name, mimeType and uploadedFile are all final. You might have to resort to using temporary local variables for storing final values on some occasions.

Ubuntu on EEE PC

2009-06-04T12:19:00.000+03:00

A few weeks ago I set out to upgrade Ubuntu on my EEE PC 901 from Intrepid to Jaunty. It was an, um, interesting experience, so I'm jotting down these notes for next time. Upgrading from Update Manager was not an option, because my root partition was not big enough to host the upgrade. The 901 has a 4GB SSD that I'm using as root and an 8GB (or 12GB) SSD that I've put /home on. It turns out that 4GB is OK for running a modern operating system, but a little short for updating it.

The usual procedure for upgrading is to use the official ISO images to boot from. As always you need a USB stick, or an external optical drive. I settled for the former, as you can see below.

The good news with Jaunty is that a special netbook edition is available for download. With previous editions you had to start with the desktop edition and do lots of tweaking, to arrive at a satisfactory result. With the netbook edition however things were unbelievably smooth. Everything worked out of the box with my EEE PC. Well, almost everything, but I will get to that in a while.

The first step after downloading the ISO (for the security paranoid at least) is verifying the download:


$ md5sum ubuntu-9.04-netbook-remix-i386.img
8f921e001aebc3e98e8e8e7d29ee1dd4  ubuntu-9.04-netbook-remix-i386.img

Then, we can write the image to the disk. I used Ubuntu ImageWriter (package usb-imagewriter) from another Ubuntu PC:

That went very smooth. Next step was booting the EEE PC with the newly formatted USB disk. Hitting ESC on boot, allows one to change the boot drive sequence. If this is the first time you are replacing the original OS, you probably need to disable Boot Booster from the BIOS first.

The installation process is quite simple. The only important thing was partitioning in my case, since I wanted to preserve my /home partition intact. Last time I installed Ubuntu I went with ext2 filesystems, trying to squeeze as much life from my solid state disks as I could. After doing some more research on the subject, I decided that I'll probably throw away my EEE long before the disks begin to wear out. Either that, or I'll throw an SD card in the expansion slot. Life is full of choices, particularly on the $300 range.

So I selected manual partitioning and chose ext3 for my first SSD. This is a screenshot from my first attempt. No, actually it was my second attempt. I screwed up the first one. Turns out I screwed up this one, too. See if you can spot why.

It may not be that obvious, but swap should not be on the first disk. If you are using swap at all (and I know many suggest that you don't), it would be a better choice to put it on the second disk. With 1GB of RAM, swap should be at least 1GB, which leaves only 3GB for Linux. The default installation needs something north of 2GB, so that leaves less than a gig to install software you use. Unless of course you are planning to play symlink tricks, in which case I should probably offer my condolences. Anyway, I ended up chopping 1GB with gparted from the second disk and redoing the installation again, but fortunately you won't have to.

What you should definitely do however, is put aside 16MB of space in the first disk for enabling Boot Booster in the BIOS again. This is an ASUS feature that persists the BIOS configuration on the disk, shaving a couple of seconds on each boot. It may not seem much, but think about how many times you are going to boot this thing. If you add them up, it starts to matter. Boot Booster needs only 8MB of disk to function, but I found that the installer's partition editor would only allocate one cylinder for 8MB, while we need two. Using 16MB did the right thing.

The regular customization I always do is to put /tmp and /var/tmp on tmpfs, so that many common operations don't touch the disks at all. Put the following lines in fstab to do that:


tmpfs        /tmp        tmpfs    defaults,noexec,nosuid        0       0
tmpfs        /var/tmp    tmpfs    defaults,noexec,nosuid        0       0

After editing /etc/fstab the proper way to remount these file systems is by following these steps:

Log out of the Ubuntu desktop in order to minimize the number of open temp files.
Press CTRL-ALT-F1 on your keyboard. You should see a login prompt. Login with your usual user/pass.
Clear out the files in the existing /tmp and /var/tmp directories.
sudo rm -rf /tmp/* /var/tmp/*
Mount the filesystems using tmpfs.
sudo mount /tmp
sudo mount /var/tmp
Reboot the EEE PC.
sudo shutdown -r now

Next, from the boot-as-fast-as-you-can department comes another tweak. In /boot/grub/menu.lst set "timeout 0" in order to skip the kernel selection delay. If I ever need to boot a different kernel, I'll change it back to something larger, thankyouverymuch.

Log out of the Ubuntu desktop. (Yes.. I said log out. This will go a long way to ensuring the fewest number of temp files are currently open.)
Press CTRL-ALT-F1 on your keyboard. You should see a login prompt. Login with your usual user/pass.
Clear out the files in the existing /tmp and /var/tmp directories.
sudo rm -rf /tmp/* /var/tmp/*
Mount the filesystems using tmpfs.
sudo mount /tmp
sudo mount /var/tmp
Reboot the EEE PC.
sudo shutdown -r now

Now in order to enable back Boot Booster follow these instructions for configuring the small partition we created earlier. When you reboot make sure to enable the option in the BIOS (it will be hidden when the proper partition isn't found).

Now you are all set. At least if you have no use for WPA-secured WiFi. Because if you do, you'll hit a bug with the included rt2860 driver. The details aren't that interesting, but the available options are two: either use the array.org kernel or patch the driver. The former solution used to be necessary with Intrepid, since lots of things were not working with the default kernel. However after trying out both, I don't see much need for Adam's kernel in Jaunty, unless you are terrified when encountering patch, make and their ilk. Since I'm on first name basis with them, I went with the second option, hoping that a future kernel update will bring along the fixes.

All in all, the process this time around was very easy. Almost easy peasy. My EEE PC with Ubuntu Netbook Remix makes me almost forget that I'm not on my beloved Mac.

Almost.

An introduction to the GSS API

2009-04-23T12:10:00.008+02:00

Application Programming Interface or API design is one of my favorite topics in programming, probably because it is both a science and an art. It is a science because there are widely accepted principles about how to design an API, but at the same time applying these principles within the constraints of a given programming language, requires the finesse of an experienced practicioner. Therefore it is with great pleasure that I'll try to explain the ins and outs of the GSS REST-like API, as promised before. As I've already mentioned, GSS is both the name of the source code project in Google Code, as well as the GRNET-sponsored service for the Greek research and academic network (although, it's official name after leaving the beta stage will be Pithos). Since anyone can use the open-source code to setup a GSS service, in this post I'll use generic examples, so anyone writing a client for the GRNET service should modify them accordingly.

When developing an application in a particular programming language, we are used to thinking about the APIs presented to us by the various libraries, which are invariably specified in that same language. For instance, for communicating with an HTTP server from a Java program we might use the HttpClient library API. This library presents a set of Java classes, interfaces and methods for interacting with web servers. These classes hide the underlying complexity of making the low-level HTTP protocol operations, allowing our mental process to remain constantly in a Java world. We could however interact with a web server without such a library, opting to implement the HTTP protocol interactions ourselves instead. Unfortunately, there is no such higher-level library for GSS yet, wrapping the low-level HTTP communications. Therefore this post will present a low-level API, in the sense that one has to make direct HTTP calls in his chosen programming language. The good news is that the following discussion is useful for programmers with any background, since there is support for the ubiquitus HTTP protocol in every modern programming language.

A RESTful API models its entities as resources. Resources are identified by Uniform Resource Identifiers, or URIs. There are four kinds of resources in GSS: files, folders, users & groups. These resources have a number of properties that contain various attributes. The API models these entities and their properties in the JSON format. There is also a fifth entity that is not modeled as a resource, but is important enough to warrant special mention: permissions.

Users · Users are the entities that represent the actual users of the system. They are used to login to the service and separate namespaces of files and folders. User entities have attributes like full name, e-mail, username, authentication token, creation/modification times, groups, etc. The URI of a user with username paul would be:

http://host.domain/gss/rest/paul/

The JSON representation of this user would be something like this:

{
    "name": "Paul Smith",
    "username": "paul",
    "email": "paul@gmail.com",
    "files": "http://hostname.domain/gss/rest/paul/files",
    "trash": "http://hostname.domain/gss/rest/paul/trash",
    "shared": "http://hostname.domain/gss/rest/paul/shared",
    "others": "http://hostname.domain/gss/rest/paul/others",
    "tags": "http://hostname.domain/gss/rest/paul/tags",
    "groups": "http://hostname.domain/gss/rest/paul/groups",
    "creationDate": 1223372769275,
    "modificationDate": 1223372769275,
    "quota": {
        "totalFiles": 7,
        "totalBytes": 429330,
        "bytesRemaining": 10736988910
    }
}

Groups · Groups are entities used to organize users for easier sharing of files and folders among peers. They can be used to facilitate sharing files to multiple users at once. Groups belong to the user who created them and cannot be shared. The URI of a group named work created by the user with username paul would be:

http://host.domain/gss/rest/paul/groups/work

The JSON representation of this group would be something like this:


[
    "http://hostname.domain/gss/rest/paul/groups/work/tom",
    "http://hostname.domain/gss/rest/paul/groups/work/jim",
    "http://hostname.domain/gss/rest/paul/groups/work/mary"
]

Files · Files are the most basic resources in GSS. They represent actual operating system files from the client's computer that have been augmented with extra metadata for storage, retrieval and sharing purposes. Familiar metadata from modern file systems are also maintained in GSS, like file name, creation/modification times, creator, modifier, tags, permissions, etc. Furthermore, files can be versioned in GSS. Updating versioned files retains the previous versions, while updating an unversioned file replaces irrevocably the old file contents. The URI of a file named doc.txt located in the root folder of the user with username paul would be:

http://host.domain/gss/rest/paul/files/doc.txt

The JSON representation of the metadata in this file would be something like this:


{
   "name": "doc.txt",
   "creationDate": 1232449958563,
   "createdBy": "paul",
   "readForAll": true,
   "modifiedBy": "paul",
   "owner": "paul",
   "modificationDate": 1232449944444,
   "deleted": false,
   "versioned": true,
   "version": 1,
   "size": 802,
   "content": "text/plain",
   "uri": "http://hostname.domain/gss/rest/paul/files/doc.txt",
   "folder": {
       "uri": "http://hostname/gss/rest/aaitest@uth.gr/files/",
       "name": "Paul Smith"
   },
   "path": "/",
   "tags": [
       "work",
       "personal"
   ],
   "permissions": [
       {
           "modifyACL": true,
           "write": true,
           "read": true,
           "user": "paul"
       },
       {
           "modifyACL": false,
           "write": true,
           "read": true,
           "group": "work"
       }
   ]
}

Folders · Folders are resources that are used for grouping files. They represent the file system concept of folders or directories and can be used to mirror a client's computer file system on GSS. Familiar metadata from modern file systems are also maintained in GSS, like folder name, creation/modification times, creator, modifier, permissions, etc. The URI of a folder named documents located in the root folder of the user with username paul would be:

http://host.domain/gss/rest/paul/files/documents

The JSON representation of this folder would be something like this:


{
   "name": "documents",
   "owner": "paul",
   "deleted": false,
   "createdBy": "paul",
   "creationDate": 1223372795825,
   "modifiedBy": "paul",
   "modificationDate": 1223372795825,
   "parent": {
       "uri": "http://hostname.domain/gss/rest/paul/files/",
       "name": "Paul Smith"
   },
   "files": [
       {
           "name": "notes.txt",
           "owner": "paul",
           "creationDate": 1233758218866,
           "deleted":false,
           "size":4567,
           "content": "text/plain",
           "version": 1,
           "uri": "http://hostname.domain/gss/rest/paul/files/documents/notes.txt",
           "folder": {
               "uri": "http://hostname.domain/gss/rest/paul/files/documents/",
               "name": "documents"
           },
           "path": "/documents/"
       }
   ],
   "folders": [],
   "permissions": [
       {
           "modifyACL": true,
           "write": true,
           "read": true,
           "user": "paul"
       },
       {
           "modifyACL": false,
           "write": true,
           "read": true,
           "group": "work"
       }
   ]
}

Working with these resources is accomplished by sending HTTP protocol requests to the resource URI with GET, HEAD, DELETE, POST, PUT methods. GET requests retrieve the resource representation, either the file contents, or the JSON representations for the resources specified above. HEAD requests for files return just the metadata of the file and DELETE requests remove the resource from the system. PUT requests upload files to the system from the client, while POST requests perform various modifications to the resources, like renaming, moving, copying, moving files to the trash, restoring them from the trash, creating folders and more. The operations are numerous and I hope to cover them in more detail in a future post.

One important aspect of every RESTful API is the use of URIs to allow the client to maintain a stateful conversation. For example, fetching the user URI would provide the files URI for fetching the available files and folders. Fetching the files URI would in turn return the URIs for the particular files and folders contained in the root folder (along with other folder properties). Returning to the parent of the current folder would entail following the URI contained in the parent property. This mechanism removes the state handling from the server and puts the burden on the client, providing excellent scalability for the service. Furthermore, since the URIs are treated opaquely by the client, the API allows client reuse across server deployments. A client can target multiple GSS services, as long as they speak the same RESTful API. Moreover, links from service A can refer to resources in service B without a problem (in the same authentication domain, e.g. the same Shibboleth federation). This is the same as using a single web browser to communicate with multiple web servers, by following links among them.

Reconciling Apache Commons Configuration with JBoss 5

2009-04-20T21:01:00.008+02:00

There are many ways to configure a JavaEE application. Among the available solutions are DBMS tables, JNDI, JMX MBeans and even plain old files in a variety of formats. While we have used most of the above in various occasions, I find that plain files resonate with all types of sysadmins, when no other administrative interface is available. For such scenarios, Apache Commons Configuration is undoubtedly the best tool for the job. Recently, I came across an undocumented incompatibility when using Commons Configuration with JBoss 5 and I thought I should describe our solution for the benefit of others.

Usually we are storing our configuration files in the standard place for JBoss, which is JBOSS_HOME/server/default/conf, for the default server configuration. This has the disadvantage that is not as easy to remember as /etc in UNIX/Linux or \Program Files and \Windows in Windows systems, but it has the important advantage of being specified as a relative path in our code, making it more cross-platform without cluttering it with platform-specific if/else path resolution checks.

Commons Configuration can reload the configuration files automatically when changed in the file system, which helps prolong the server uptime. However JBoss 5 has introduced the concept of a virtual file system that caches all file system accesses performed through the context class loaders using relative paths. Unfortunately this generates resource URLs in the form vfsfile:foo.properties that Commons Configuration does not know how to deal with. Fixing this requires extending FileChangedReloadingStrategy, like we do in gss. Alternatively, one could patch Commons Configuration with the following change and use the standard FileChangedReloadingStrategy unchanged:


Index: FileChangedReloadingStrategy.java
===================================================================
--- FileChangedReloadingStrategy.java    (revision 764760)
+++ FileChangedReloadingStrategy.java    (working copy)
@@ -46,6 +46,9 @@
  /** Constant for the jar URL protocol.*/
  private static final String JAR_PROTOCOL = "jar";

+    /** Constant for the JBoss MC VFSFile URL protocol.*/
+    private static final String VFSFILE_PROTOCOL = "vfsfile";
+
  /** Constant for the default refresh delay.*/
  private static final int DEFAULT_REFRESH_DELAY = 5000;

@@ -161,7 +164,8 @@

  /**
   * Helper method for transforming a URL into a file object. This method
-     * handles file: and jar: URLs.
+     * handles file: and jar: URLs, as well as JBoss VFS-specific vfsfile:
+     * URLs.
   *
   * @param url the URL to be converted
   * @return the resulting file or null 
@@ -181,6 +185,18 @@
              return null;
          }
      }
+        else if (VFSFILE_PROTOCOL.equals(url.getProtocol()))
+        {
+            String path = url.getPath();
+            try
+            {
+                return ConfigurationUtils.fileFromURL(new URL("file:" + path));
+            }
+            catch (MalformedURLException mex)
+            {
+                return null;
+            }
+        }
      else
      {
          return ConfigurationUtils.fileFromURL(url);

Neither of these solutions covers the case of storing configuration files in zip or jar containers, but since it is something I haven't found a use for yet, I can't test a fix for it. If anyone is interested in such a use case, I'd advise extending FileChangedReloadingStrategy, combining the logic in jar: and vfsfile: URL handling.

GSS architecture

2009-04-13T22:38:00.006+02:00

Handling a large number of concurrent connections requires many servers. Not only because scaling vertically (throwing bigger hardware at the problem) is very costly, but also because even if an application can be designed to scale vertically, the underlying stack probably can not. Java applications for instance, like GSS, run on the JVM and although the latter is an excellent piece of engineering, using huge amounts of heap is not something it's tuned for. Big Iron servers with many cores and 20+ GB of RAM are usually running more than one JVM, since garbage collection is not all that efficient with huge heaps. And since running application instances with a 4-8 GB heap size can be done with cheap off-the-shelf hardware, why spend big bucks on Big Iron?

So having a large number of servers is a sane choice, but brings it's own set of problems. Unless one partitions users to servers (having all requests a particular user makes be delivered to the same server), all servers must have a consistent view of the system data, in order to deliver meaningful results. Assigning user requests to particular servers, usually requires expensive application layer load-balancers or customized application code on each server, so it would rarely be your first option. Having all servers work on the same data is a more tractable problem, since it can be solved by having the application state being replicated among server nodes. Usually, only a small part of the application state needs to be replicated, for each user, that is the part which concerns his current session. But even though session clustering solutions have been a well studied field and implementations abound, having no session to replicate is an even better option.

For GSS we have implemented a stateless architecture for the core server, that should provide us with good scalability in a very cost-effective manner. The most important part in this architecture is the REST-like API that moves part of the responsibility for session state maintenance to the client applications, effectively distributing the system load to more systems than the available server pool. Furthermore, client requests can be authenticated without requiring an SSL/TLS transport layer (even though it can be used if extra privacy is required), which would entail higher load on the application servers or require expensive load balancers. In the server side, API requests are being handled by servlets that enlist the services of stateless session beans, for easier transaction management. Our persistence story so far is JPA with a DBMS backing, plus high-speed SAN storage for the file contents. If or when this becomes a bottleneck, we have various contingency plans, depending on the actual characteristics of the load that will be observed.

The above image depicts the path user requests will travel along, while various nodes interact in order to serve them. A key element in this diagram is the number of different servers that can be found, effectively specializing in their own particular domain. Although the system can be deployed on a single physical server (and regularly is, for development and testing), consisting of a number of standalone sub-services instead of a big monolithic service is a boon to scalability.

This high-level overview of the GSS architecture should help those interested to find their way around the open-source codebase and explain some of the design decisions. But the most interesting part from a user's point of view would be the REST-like API, that allows one to take advantage of the service for scratching his own itch.

So that will be the subject of my next post.

Introducing GSS

2009-04-08T16:40:00.000+02:00

During my recent work-induced blog hiatus, I've been working on a new software system, called GSS. I've been more than enjoying the ride so far and since we have released the code as open-source, I thought discussing some of the experience I've gained might be interesting to others as well.

GSS is a network, er grid, er I mean cloud service, for providing access to a file system on a remote storage space. It is the name of both a service (currently in beta) for the Greek research and academic community and the open source software used for it, that can also be used by others for deploying such services. It is similar in some ways to services like iDrive, DropBox and drop.io, but it can also be regarded as a more high-level Amazon S3. Its purpose is to let the desktop computer's file system meet the cloud. The familiar file system metaphors of files and folders are used to store information in a remote storage space, that can be accessed from a variety of user and system interfaces, from any place in the world that has an Internet connection. All usual file manager operations are supported and users can share their files with selected other users or groups, or even make them public. Currently there are four user interfaces available, a web-based application, a desktop client, a WebDAV interface and an iPhone web application, in various stages of development. Underlying these user interfaces is a common REST-like API that can be used to extend the service in new, unanticipated ways.

The main focus of this service was to provide the users of the Greek research and academic community with a free, large storage space that can be used to store, access, backup and share their work, from as many computer systems as they want. Since the available user base is close to half a million (although the expected users of the service are projected to the low ten thousands), we needed a scalable system, that would be able to accommodate high network traffic and a high storage capacity at the same time. A Java Enterprise Edition server coupled with a GWT-based web client and a stateless architecture were our solution. In future posts I will describe the system architecture with all the gory details. The exposed virtual file system features file versioning, trash bin support, access control lists, tagging, full text search and more.

All of these features are presented through an API for third party developers to create scripts, applications or even full blown services that will fulfill their own particular needs or serve other niches. This API has a REST-like design and though it will probably fail a formal RESTful definition, it sports many of the advantages of such architectures:

system resources such as users, groups, files and folders are represented by URIs

GET, HEAD, POST, PUT and DELETE methods on resources have the expected semantics

HTTP caching is explicitly supported via Last-Modified, ETag & If-* headers

resource formats for everything besides files are simple JSON representations

only authenticated requests are allowed, except for public resources

Users are authenticated through the GRNET Shibboleth infrastructure. User passwords are never transmitted to the GSS service. Instead GSS-issued authentication tokens are used by both client and server to sign the API requests after the initial user login. SSL transport can provide even stronger privacy guarantees, but it is not required, nor enabled by default.

The GSS code base is GPL-licensed and therefore anyone can use it as a starting point to implement his own file storage service. We have yet to provide binary downloads, due to the various dependencies, but the build instructions should be enough to get someone started. We are always interested in source or documentation patches, of course (did I mention it's open source?). Most importantly, the REST API will ensure that clients developed for one such service can be reused for every other one.

I will have much more to say about the API in a future post. In the meantime you can peruse the code and documentation, or even try it out yourself. I'd be very interested in any comments you might have.

Applications meme

2009-03-14T15:00:00.001+02:00

It's been a while since my last post, but I've been wanting to write about the interesting stuff that I've been working on. Unfortunately it took longer than anticipated, but now we're almost done and I'll be posting soon with the gory details. However, since I got tagged, I suppose I can throw in a not-so-interesting post, about my Linux application preferences.

Which desktop manager do you use more often ?
Gnome. It's the only one that reminds me of OS X somewhat (appearance- and functionality-wise).
Which desktop application you would not like to see implemented again on linux? And why?
A desktop manager. If you are going to provide yet another incarnation of a 30-year old idea, please don't bother. The concept is pretty well understood: icons/workspace/file manager/drag'n'drop/mouse/menus/launcher/whatever, it's been done before. Like a million times. If you want to be useful, try to fix some of the shortcomings of the existing ones (notification mechanisms, translucent effects, faster access patterns, application integration, etc.).
Which desktop application you definitely would like to see implemented on linux? Describe it briefly or point out to a similar application.
Adobe Premiere, or iMovie, or any other semi-decent video editing software. This was the main reason I bought a Mac.
Write the name of the last project (not the very best, the last!) that made you wish to thank their developers so you can thank them now!
Git. You guys rock! (And if you improve on the UI, you will jazz!)
(Optional) Link the blogs of 1-3 people you’d like to take part to this meme. (no more than three). you can skip this question if you like.
Let's hear from chstath, ckotso and keramida (who may s/Linux/FreeBSD/ at his discretion).

Programming for fun

2008-10-07T23:30:00.000+02:00

I've been in a hiatus from blogging lately, mainly because free time has been in short supply. It's not just the very busy days at work, nights have been even busier. Instead of churning out long blog posts, I've been banging on code instead, day and night. Unfortunately I only get paid for the code I write during the day, but surprisingly (or not) I enjoy the code I write at night the most.

I have written code for banks, multinational corporations, the kind of places where the phrase "professional dress code" is not used strictly in jokes. I've helped build software that was resilient, scalable, secure and intuitive, although granted, not all those at the same time. This was mainly Custom Enterprise Software, or to put it more plainly, Software That Runs On Computers At Work. So I thought I'd try a different angle in my spare time: write software that people use at work, or at home. Or at a coffee shop. Or at the airport. Or wherever they feel like using a computer, anyway. This is sometimes called End-User Software, or to put it in layman's terms, Software People May Actually Like. After trying out some ideas that didn't seem worthwhile, I discovered that some of my buddies had similar thoughts. This is how FireStatus was born.

FireStatus is a Firefox extension that aims to be a swiss army knife for dealing with various social networks, right from your browser, without visiting any particular website. Facebook, Twitter and FriendFeed are currently supported, but more are in the pipeline. For starters, it allows you to simultaneously update your status to all or some of these services, so that all your friends see it, no matter what they are using. The notion of a status that is occasionally updated is familiar to Twitter and Facebook users, since the text field that asks 'What are you doing?' is prominent in their user pages. FriendFeed does not have the notion of a status, but its users can post short (or long) messages, just like Twitter's. FriendFeed also allows posting links to web pages, accompanied with a short description, something that many Twitter and Facebook users have been doing by constructing status updates that start with the description text and are followed by an appended link, usually shortened.

FireStatus can ease the task of posting these messages or status updates, by being always available, instead of needing to have the service pages open and without a large memory footprint, like other similar applications, since it takes advantage of the fact that most people nowadays always have one browser open. I know I do. Come to think of it, I don't know anyone who uses computers that does not keep a browser running most of the time. At least untile the darned thing crashes, which fortunately is something I haven't experienced in Firefox for quite some time.

Posting a message or status update

Clicking on the FireStatus icon, pops up a small toolbar window just above the status bar and below the window document. It is similar to the Firefox findbar that pops up when one searches for text in a page, albeit slightly larger. That was a deliberate design decision that aimed to imitate the success in the usability of the findbar. Lots of little details like this one have been carefully thought out and occasionally debated at length among the team and our beta testers:

Enabling the spell checker in the status message field, for catching those typos when hastily typing a message.
Having the URL inclusion unchecked by default, while the shortening checked, since most posts do not include URLs (and it might be embarrassing if done inadvertently), but those that do usually want them short.
Adding a character counter to help guard against the maximum message length imposed by some services (e.g. Twitter).
Tightly placing the available services in a way that allows for an unambiguous selection.
Using the Escape and Enter keys as shortcuts for canceling and sending the update respectively.

As one reviewer in addons.mozilla.org so eloquently put it, "the simplicity & minimal design are the key components here".

Posting is not the whole story, though. These services provide continuous streams of updates from friends that we need to monitor. Sifting through the updates in every service, while very useful, becomes tedious after a while. It is like making a mental note to check your e-mail every once in a while for new messages. It's one way to do it for sure, but mail notifiers have been around for ages and provide more of a "live" experience.

FireStatus imitates their success in making e-mail conversations "alive", for conversations in the social network space. Every time a new incoming update is received, a notification popup appears, so I can continue using whatever application I was using, but being instantly aware of the news. Twitter notifications contain the name and the picture of the message author as well as the message sent. Low-priority or uninteresting messages can be just glanced at and then automatically dismissed, while important ones can be clicked on, so that the message can be viewed in the browser. FriendFeed notifications can be of various flavors, since FriendFeed aggregates updates from a large variety of online services.

Therefore FriendFeed updates display the service icon along with the author's name for quick identification of the type of update. Clicking on an interesting update opens a browser window to the link contained in the message. Facebook notifications currently contain new message counts, pokes and shares, but work for getting more data is underway.

Although the notifications appear instantaneous, as with e-mail notifications there is a polling process involved underneath. The polling frequency for each service can be separately tuned in the preferences window. The preference window can be opened by right-clicking on the FireStatus icon and selecting the appropriate option in the popup menu. Along with the time intervals between polling for updates, one can enable posting and/or receiving updates for the various services, as well as the authentication credentials where appropriate. For Facebook the user logs in in a Facebook-supplied browser window that pops up, similar to regular Facebook use. No credentials need to be stored separately by FireStatus. For FriendFeed the username and remote key have to be stored locally in the extension preferences. The remote key is not the same as the user password, but is provided by FriendFeed specifically for use by third-party applications, like ours. It can be found by clicking on the link displayed in the explanatory note. Twitter can work either way. If a username and password are entered in the preference window, they will be stored locally in the extension preferences and used for subsequent authentication. If, on the other hand, the fields are blank, FireStatus will consult the browser's Password Manager for any stored Twitter credentials. This will probably cause the master password dialog to pop up once, but the credentials will remain stored only in the Password Manager.

FireStatus preferences

Since we built this thing for our own personal use, pleasure and self-punishment, and as we don't plan to ever get rich out of it, we have released it from the beginning as open-source software through the permissive BSD license. You are most welcome to come by the project's home at Google Code, study the code, fix a bug (or two), help with the (embarrassingly outdated) documentation, contribute a review (otherwise it will remain an experimental extension forever) or just chat with us at the mailing list. Heck, if you feel like it, you can grab the whole code and use it in your own lucrative business and you don't owe us a dime.

With the shape the world economy currently is in, you probably couldn't afford it, anyway.