Installing l10n packages for all software in resulting rootfs

This one is rarely used — I know one company which makes use of it. How does it works? It is called after root filesystem is populated with packages and goes one by one and install “-locale-LANG” for each required language. As you may expect it makes whole process much, much longer. Activation is easy:

ROOTFS_POSTINSTALL_COMMAND += "install_all_locales; "

Code is available only for building from IPKG packages and is stored in rootfs_ipk.bbclass file. So far that code is present only in Poky Linux.

Building AUTOREV packages over slow link

How many times you had a situation when you did a build and many components used SRCREV="${AUTOREV}" setting? Parsing can take eons then…

But there is a solution. It will cache values of all SRCREV variables so just one parse will be long — next one use cached values. Of course you lose automatic revisions but instead you have fast parsing time which is blessing when most of your work is fixing build problems. Activation is easy:

BB_SRCREV_POLICY = "cache"

Save disc space by not generating GIT tarballs in DL_DIR

How many of you readers share your DL_DIR (directory where OE stores all fetched sources) with others? For those who answered “not me” there is a way to gain some space by not generating tarballs with GIT repositories.

By default BitBake clones GIT tree and checkouts it. Then 2 archives are generated: one with just “.git” directory and second with sources — the one which is used in do_unpack task. So why waste space for storing first one? We have it somewhere in DL_DIR/git/ for next time… Activation is easy:

BB_GENERATE_MIRROR_TARBALLS = "0"

More tricks in next posts.

Related posts:

1. My OpenEmbedded environment II
2. My OpenEmbedded enviroment
3. 30 months of OpenEmbedded and me

At start road was selected quite properly, the problems started later. Many times it was recalculating route (the funniest was when I stopped at traffic lights it recalculated route and then made it again after driving for 5 meters). Directions changes information were also distracting: “on 200m go left” then “go right” when I reached point. Or “go left,right” when it was straight road or when I had to turn right at crossing.

I have to admit that after two tries I totally gave up on driving with it and switched back to TomTom device. But there is one good thing in Ovi Maps — it works quite good for walk navigation. I used it in many cities abroad and it handled. I hope that Maemo5 version will work at least that good as Symbian one.

Related posts:

1. Online map services and routing
2. Car navigation with N810
3. Nokia N900 discount

Ok, so let’s test it. As official release notes mentioned Polish translation I switched to it. What I saw was disaster… And the problem is that latest firmware has same version available. Examples:

Note: I did system update by using “apt-get update;apt-get upgrade” followed by “fakeroot dpkg --configure -a” (because sbox root ‘emulation’ was too weak for “libosso-abook”). As a result translation “6.0+r7250+0m5″ were installed (”calendar-l10n-plpl” package). Scary, isn’t it?

Next thing which I looked at was instant messaging — which I was not able to use since there was no Jabber support. It has ICQ, Gadu-Gadu, MSN, Sametime, Salut, QQ, Groupwise, Google Talk, AIM, Yahoo but no Jabber. I even installed “telepathy-extras” but that added me yet another MSN and still no Jabber… Probably other package needs to be installed but why it has to be? OK, maybe SDK image does not contain same packages as device one but why it supports such poor thing like Gadu-Gadu out-of-box?

Contacts app is improved when compared to Maemo4 one — finally something closer to application which I described some time ago in other post. Has some bugs (I can not enter “+48123456789″ as number because only digits are available (and ‘p’ for internal numbers).

Looks like there will be a long list of things to report as bugs when/if I will get my N900 (so far it is somewhere in deep nowhere). Why not reporting them now? Simple — I want to be able to check them on device itself.

Other things I will check next time. Now it is time to sleep.

Related posts:

1. Nokia N800 emulation
2. Nokia N900 discount
3. Localizing Maemo

Board is much better then EDB9301 which I used so far for EP93xx toolchain tests. What is on board:

• EP9307 CPU
• 64MB ram
• 8MB NOR flash
• MMC slot (connected over SPI so ~250KB/s max)
• 2 USB host ports
• VGA out port — XGA 8bit or SVGA 8/16/24bit
• serial port in RJ-45 instead of standard DB9 (but cable is in package)
• audio in/out jacks
• many connectors with different signals — will have to check schematics for that.

By default board boots into Debian ‘lenny’ system stored on 4GB SDHC card. But there are problems with it as this is MMC over SPI so speed is very limited (about 250KB/s only) and it time outs quite often so I plan to move to USB stick during next days.

Next step will be adding it into OpenEmbedded and running Ångström as base distribution.

BTW — how did I got it at all? That’s due my recent work on merging EP93xx support into OE — I was asked do I want developer board with this processor.

Related posts:

1. EP93xx fight continued…
2. EDB9301 hacking

It got over 50 changes during 1.5 year of development. Most of them were done by me as part of my work for Bug Labs company and their BUG Linux distribution. As policy of handling fixes requires to make them also in development branch it was more then just make a fix for ‘pinky’ — I also had to take care of ‘elroy’ (which had to be next stable version) and ‘master’. Some time ago support for ‘elroy’ was dropped anyway.

If you look at ‘git log’ output you will notice few authors other then just me. Thats because if fix was present in other sources such like Poky ‘master’ or OpenEmbedded I cherrypicked it and adapted to make it apply with keeping original author credits.

Did I add something new into it? Yes, few things were added:

• SPLASH support in task-poky so you can use own bootsplash tool instead of psplash
• warning for ‘/proc/sys/vm/mmap_min_addr’ not being set to 0 (which would break qemu)
• Python 2.6 compatibility
• BP/BPN variables which were used in Jalimo repository which we use as one of overlays
• automatic resizing of ext2/3 images if rootfs do not fit in default size

What will future bring? I hope that new stable branch for Poky will be created in next few months so developers will be able to switch. I know that some companies did a move from ‘pinky’ to ‘master’ (or snapshot of it + own changes). We at Bug Labs are moving into OpenEmbedded ’stable/2009′ as we need newer software and want some functionality which is not present in ‘pinky’.

Related posts:

1. Back to Poky Linux development
2. Contributing to Poky is not hard
3. Five years with OpenEmbedded

I travelled by Berlin and Stansted due to fact that at those days there are no flights Londyn — Szczecin on Friday and Sunday. That gave opportunity to meet Henning and Robert in Berlin and discuss misc things during travel.

Friday evening was spent in Red Lion pub in Histon. Nice, tasty English beers and interesting place. Both ways with taxi because of raining and price was comparable to public transport.

On Saturday we walked to OEDEM place using Phil’s notes and got there with just one short cut missed (we found it next day).

What was discussed during meeting? Many things, we had also OE e.V. meeting during which we voted for few new members, chosen new board, decided on sponsoring and selected Robert Schuster for PR guy.

So what was on topic:

• OE e.V. meeting (minutes)
• Establishment of technical steering committee (summary)
• OE and Poky (summary)
• Software development for OE derived distributions (summary)
• Learning to love distro flags, or, What’s a distro for? (summary)
• State of the art in package management (summary)
• Splitting the recipes tree (summary)
• Future plans for stable branch(es) (summary)
• Bugtracker Discussion (summary)
• Hosting arrangements (summary)
• Death to checksums.ini? (summary)
• BitBake Future Roadmap (summary)
• OE Core Changes (summary)

Each entry was already summarized so I provided links for those who did not read OpenEmbedded development mailing list. Discussion was hot, many subjects had different opinions from audience and it was great.

As usual it was nice to meet friends from OE in person and for few of them connect face to name. And again I did not had any spare time to look around the city of OEDEM — too many things in small amount of time. But maybe another time :)

Related posts:

1. One step closer to OEDEM
2. OEDEM 2009
3. OEDEM 2006

Plan to write review after first days of use so stay tuned.

Related posts:

1. Nokia N900 discount
2. My Neo1973 arrived
3. My opinion on next Nokia tablet

But that does not mean that I ended supporting this build system. I started contract work for BugLabs company and they are using Poky ‘pinky’ release. During that time I fixed some bugs (mostly by backporting fixes from Poky ‘master’ or OpenEmbedded) but most of time did not touched main development branch.

Until now — I have customer which uses main branch for own development. So looks like I will provide some new code for development branch during next weeks.

Related posts:

1. Poky Linux 3.1.2 released
2. Five years with OpenEmbedded
3. HaeRWu created

It got several updates today:

• Linux kernel 2.6.32-rc4 from Open-RD git tree
• U-Boot 2009.08 from mainline Marvell branch
• Samba 3.3.9 for sharing data over network
• NFS kernel daemon for sharing storage for my devices

Yes, two storage serving daemons. Samba will provide access to multimedia data when NFS will be more used for developer board root file systems and such.

Now hard drive is working connected via ESATA but after copying it will be connected to Sheevaplug instead of my workstation.

Just to cut questions — I bought Welland ME-752H enclosure. It allows to use Serial ATA 3.5″ drives and connect them by USB or ESATA interfaces. The bonus part is 2 port USB hub so I will not be out of USB ports in Sheeva ;)

Related posts:

1. Sheevaplug arrived
2. New interesting devices
3. Does Vortex86sx based devices are worth something?

The whole event took place in Grenoble, France on 15-16 October. I was there as a guest of ST-Ericsson company and had one talk there.

There were just 3 tracks so it was much easier to choose which talks to attend. I was on:

• Jon Masters “Porting Linux” — interesting and it was nice to see the person which do LKML summary podcast
• Grégory Clement “How We Got a 3D Application Booting in 5 Seconds Under Linux” — author told about few tricks they tried in a bit boring way
• Sascha Hauer, Marc Kleine-Budde “U-Boot-v2″ — project looks nice and I hope that my devices will migrate to this one day
• Cedric Hombourger “Why OpenEmbedded Proved a Good Foundation for MontaVista” — lot of people attended, interesting talk it was. Nice part was when Cedric told that they checked all of currently available free build systems and chosen our one.
• Samuel Ortiz “Linux Wifi Solutions for Mobile Platforms” — lot of technical informations but now I know more about Linux WLAN stack that I did
• Matt Porter “Mythbusters: Android” — nice talk about this system and which parts of it needs lot of work to be usable on other devices then just phones. I was also interested in it as I had Android on Nokia N810 during ELC-E but finally decided to go back to Maemo.
• Alessandro Rubini “Use of the Fast IRQ (FIQ) in ARM-Linux” — again lot of technical informations but I really like how Alessandro do his talks.
• Alex de Vries “Technical Features and Components of Open Source Build Systems” — nice summary of what open embedded distributions lacks and what needs to be done. Was fun to listen how Alex tries to not say “open embedded distributions” too often ;)
• Pierre Pronchery “Hackable Devices: The New Possibilities of Open Hardware”
• Michael Opdenacker “Small Business” — most of small companies which attended have OpenEmbedded services in a list of what they do

I planned to attend few more but decided that Friday will be a good time to see Grenoble in other places then just hotel and conference centre.

What was a nice surprise for me was the amount of people familiar with OpenEmbedded. Most of people which I spoke with used it for misc projects. I had a talk with one guy about “stable” branches and how our view differ. For me “stable/2009″ branch is what has to be buildable all the time but can take some updates, he prefers project branches with halted development and just very important fixes related with project.

My “Hacking with OpenEmbedded” talk got some interesting questions. Mostly about Maemo5 support but I do not know too much about it’s status.

Was it worth going? Definitely yes. Met friends, got new contacts, discussed about different projects. Was good to be there. I just hope that next time it will be in a place easier to get to — I had to fly thought Amsterdam so whole trip took me about 12 hours (each way). But on return trip I had interesting talk with Ruud Derwig during flight :D

Will something from ELC-E lands in OpenEmbedded area? Few things:

• I had a talk with Atmel guys about their work on updating linux4sam website and their OpenEmbedded support. AT91 devices will get 2.6.30 kernel in OE and it will land also in “stable/2009″ branch on which they plan to base their repository (instead of old stable branch).
• my work on ST-Ericsson NHK-15 board starts to appear in official metadata. I still need to test it on board but base is added.
• one company plans to release OMAP3 based developer board soon and this will be their first use of OpenEmbedded

Related posts:

1. ST-Ericsson Community Workshop 2009
2. Poky Linux 3.1.2 released
3. ST-Ericsson Open Source Community Workshop 2009