The good news is that it’s not hard to make your locally served websites accessible, it’s just a matter of choosing the method that suits your needs. Unfortunately there’s a lot of information on this topic out there that doesn’t quite give a complete solution, and I think some of the solutions out there are based on an older version of parallels.
So, without further ado, here’s how I’ve setup Parallels 3.0 for Mac with a guest OS of Windows 2000 and a series of websites served locally on the mac via MAMP Pro.
- Change the default parallels networking setup in System Prefs - There are 2 of them, you want to change the one that includes “(en3)”. Change it from DHCP, to DHCP with manual address Enter in an address like: 192.168.2.102
- Add an alias for the domain you are working on (bottom part of MAMP window), use the IP you entered above, save changes to mamp.
- Be sure to be on shared networking in parallels
- In parallels, open the ip, plus the port in your browser, it works!
- Use a unique port for each site, and add an alias for each site in MAMP
Please don’t hesitate to leave a comment on this post about your own experiences, this is by no means exhaustive and does not account for firewall settings, and probably many other factors. I am by no means qualified to do more than just report what works for me, and would welcome expertise on the subject here.
Update
Parallels 4.0 is out now, I managed to get the setup above working there too, however there are some modifications.
- You may need to set your parallels network adapters (in system prefs, network) to “manually” configure. Instead of being labeled “(en3)” they are now labeled the more logical “host only” and “shared.” The one you want to work with is “shared”. I set my IP to “10.211.55.2″ and my subnet mask to “255.255.255.0″, the rest I kept blank.
- The default for network settings in Parallels itself has to be set back to “shared” and “shared networking” in the little web icon at the bottom of your virtual machine window (or in the virtual machine config window)
Ah, this is getting cleaner now. I’m really enjoying the scaffolding feature of Code Igniter. It’s a little set of controllers and views that are purely for providing a web interface to administrate your database tables. Yes, I can interface with the database directly (I use the Sequel Pro GUI app), but having these web interfaces auto-created just feels easier.
I’ve added on to the scaffolding by creating a little menu page so that every table can be accessed from a single starting point. It’s like an adhoc admin tool for initial data entry. Of course this is only meant for local development, the scaffolding will be removed prior to going live since it’s a wide open door to the database. Meanwhile, the open door is making things so much simpler!
]]>The last couple days I’ve had the privilege of being able to share my database schema and intentions with a real application developer. He pointed me towards some simple, yet important conventions - perhaps most critically, a logic for database table creation that included the use of relationship tables to associate bits of information with other bits of information (especially where the relationship is one to many or many to many). For example if I have a page table, and multiple slots can be on a page - it’s best to setup a page-slot relationship table that is solely used to associate page ids with slot ids and perhaps sorting numbers. This is a shift in approach that prompted to me step back and diagram my schema again. I broke out the OmniGraffle and used a Rails database stencil (it looked nice) to map my tables. I now have 10 content tables and 5 relationship tables.
This in turn reminded me how new I am to all of this - and so I’ve decided to archive my pre-existing code (it would have to be re-written anyway), and start with a fresh install of Code Igniter (CI). The complications I created previously by trying to subvert the way CI handles URLs seemed not worth-while now, especially given the benefit of doing things cleanly and with as standard a method as possible.
So I feel rejuvenated by a fresh slate, cleaner approach - but much work is needed. I do feel that this is now more “portable”, and less hackish. Lets see if I can keep it that way.
]]>So this “non-cms cms” is really just a way of layering controls into each page as its rendered (assuming proper logged-in authority.) Some aspects of the structure that make things easier (maybe):
- The system is not intended to work “out of the box” for the client. Configuration and theme building are expected of the web developer, AKA the initial site setup.
- The project/client specific stuff is cleanly separated into theme folders
- Structurally the web developer has control of how many templates and “content module types” are available when the client is adding pages and content. The ability to add new ones means tons of flexibility.
- Content modules can be added/removed and sorted in their columns. Previously added content can be duplicated elsewhere.
And some aspects that I’m worried about:
- All this flexibility will come at a cost to complexity of the final product
- With the site structure so dynamic, too much will be required from database interaction just to render the page.
- I wonder if I’ve made too many tables: pages/page templates/slots/modules/module templates/blocks/texts/images/… requiring to many queries
- I’ve subverted the Code Igniter default URL handling so that even the first portion after the domain name is reflective of the content (and not just the name of a controller) - will I regret this later?
More soon…
]]>First some “me” background:
- I am not trained in any kind of computer programming
- I’ve been involved with the architecture (mostly advisory) of many web app developments for over ten years, but haven’t directly developed anything
- My schooling is in illustration and art, but I’ve been doing some flavor of HTML/CSS for a living since an internship at my last year of school (1996)
- I am learning as I go here
I point these things out not just for context, but as a mini-disclaimer: I will likely report things that I’m doing that should never be done; hopefully I will learn from my mistakes, but please remember that I’m a novice. Corrections will be welcome with open arms.
The project
I’m making a CMS, or really I’m making a site management tool. Over these last however-many years, I’ve worked-with, helped develop, and experienced many different tools for site upkeep. There are so many options out there - desktop apps like dreamweaver and contribute, CMS portal things like Typo3 and Mambo, and smaller-focus tools like the text editing script “Nimbletext” that Arturo, and I developed (with Chris) back when we ran Twinsparc. That’s just sampling, there are hundreds of options out there, with different focuses, expected audiences, etc.
Despite all the options, I see an opportunity for a new tool, one with a specific tool->developer->client expectation. I see a need for something that allows great flexibility for the designer/developer, without the need for either the developer or the client to absorb a complicated multi-layered navigation and user interface. One of the reasons I think tools like Tumblr and Soup.io are so exciting is that they quench the thirst for a meaningful type of simplicity - the type that allows me to do what I want by mostly getting out of my way. They both provide an interface that is created to anticipate needs and be powerful, without being a product of the structures used to create them.
So many tools are really just ugly, confusing database content editors. Really I think many of them would be better off turning off the “admin” section and just having people edit database fields directly.
That’s all the time I have for this initial post, more hopefully soon. Next time I’ll talk about the framework I chose to start with: CodeIgniter, and my initial setup.
]]>All this lead me to wondering, of course, where are these implementations if compared to other languages in a computer langauge shootout? JavaScript was far behind in the October 2007 SpiderMonkey implementation, sure. Where does it come today? I’m sure it fares better than many other dynamic languages it’s compared to. Where does it compare to ActionScript 3? Where does it compare to Perl, Python or LUA?
]]>As I’m staying a couple of days longer than the event itself, I’m also interested in meetings in the margin of WDS08 - give me a ping at andreasb[at]opera.com if you want to set something up, even especially if it’s just to go for a good espresso :-)
Adobe already seem to have left their strong position in the driving seat of ECMAScript 4 in favour of getting all browser makes onboard ECMAScript Harmony instead. (Though Harmony will probably end up being called ECMAScript 4 in the end in any case…) What does these two events mean for Tamarin, ActionScript and Adobe?
]]>- JScript Microsoft Internet Explorer 8 beta 2
- 11421.6ms
- Futhark Opera 9.60 (Kestrel 10408)
- 7103.8ms
- ActionMonkey Mozilla Firefox 3.1a1pre Minefield 2008090218
- 6701.2ms
- JavaScriptCore Apple Safari 3.1.2
- 5651.2ms
- SpiderMonkey Mozilla Firefox 3.1a2pre Minefield 20080826034105 No-JIT
- 4262.0ms
- SquirrelFish Safari WebKit-r35928
- 3454.0ms
- V8 Google Chrome 0.2.149.27 (Official Build 1583)
- 2474.2ms
- TraceMonkey Mozilla Firefox 3.1b1pre Minefield 20080902174430 JIT
- 2064.4ms
This on the same hardware and Windows install as the SunSpider 0.9 ECMAScript benchmark results from just before Christmas last year, where IE7 takes all of 55609.2ms to get through the test, Ff2 takes 25343.2ms, and Opera 9.24 (Merlin 8732) takes 21327.2ms and Safari 3.0.4 takes 14479.8ms. Going from 55609.2ms on IE7 to 2064.4ms in Ff3.1 with TraceMonkey, that’s a whooping 27 times faster. It will be really interesting seeing the responses from Opera and Microsoft to these recent extraordinarily fast scripting engines.
]]>Download Opera 9.50 and read the Opera 9.50 change logs. An article about the changes directed more at web developers can be found at Dev.Opera: Opera 9.5 - the next generation of web standards.
]]>