1. Prezentacja

    Wczoraj prowadziłem wykład o OpenEmbedded w ramach cyklu wykładów organizowanych przez PLUG Poznań. Miałem zacząć o 18:00 ale uznaliśmy z organizatorem, że zaczekamy z 10 minut (na sali było 8-10 osób). Ostatecznie było około 20 osób.

    Mówiłem około godziny, po czym padło kilka pytań — ogólnie miałem wrażenie, że wiele osób nie wiedziało zbytnio o czym mówię. Nie wiem czy było to spowodowane moim wystąpieniem czy faktem, że projekt OpenEmbedded jest mało znany w Polsce.

    Po wykładzie przy kawce kolega z pracy zaskoczył mnie najbardziej mówiąc

    Nie miałem pojęcia, że OpenEmbedded jest tak dużym projektem. Myślałem, że to coś do palmtopów…

    Written by Marcin Juszkiewicz on
  2. RPM is Really Poorly Maintained

    I created openembedded-essential package for Debian derived distributions which depend on all packages required by OpenEmbedded. It took me about 5 minutes to get first version working and 15 to get it polished. Resulting package depends on needed stuff and contain only own changelog.

    Then I tried to create RPM version for all those Red Hat/Fedora/Mandriva derived distros… RPM HowTo is from 1999 year and RPM 4.4.1 refuse to work with spec from this HowTo. So I started to dig for other manuals — found one from 2002, then some others. But still no idea how to build RPM package which DOES NOT have sources and/or patches — just pure dependencies and own changelog.

    How does RPM based maintainers do their job? Thats probably a good question…

    Written by Marcin Juszkiewicz on
  3. Google Summer of Code T-Shirt arrived

    When I was checking my mailbox (snail one) I met guy from FedEx who wanted to enter building and called my flat via intercom. I asked him does package is for my fiancee or for me and because it was for me I took it from him. At home I discovered that this unexpected package was from Google with black XL T-Shirt from GSoC program.

    US sizes are bigger then Europe ones…

    Written by Marcin Juszkiewicz on
  4. OpenEmbedded documentation day

    Today was my OpenEmbedded documentation day — I wrote several parts of user-manual:

    • what is task-base and how to use it
    • how to create own distribution
    • how to add own machine into OE
    • how to create own image

    For me part about task-base is most important because most of changes related to it was made by me so more and more people are asking what is it and how to use it. Now I can point them into OpenEmbedded documentation.

    But what is this task-base at all and why we force everyone using it in OE? Let me start from beginning.

    First we had task-bootstrap and BOOTSTRAP_ variables which listed needed tools, kernel modules and other packages. We used those in machine configs and distro configs. Some of distros got them better, some not so for example distro TGB told ‘we support WPA out-of-box’ but users started to report that they have problems with WPA-TKIP when WPA-PSK works. Adding one kernel module solved it. But then other distro maintainer had to make this list again (or copy from TGB config). Now each distro which has ‘wifi’ in DISTRO_FEATURES* will get proper list of kernel modules and helper utils by default.

    As a bonus we are able to cleanup machine and distro configs from all those BOOTSTRAP_* things so now we can have something like this:

    TARGET_ARCH = "i686"
    IPKG_EXTRA_ARCHS = "x86 i386 i486 i586"
    
    PREFERRED_PROVIDER_virtual/kernel = "linux-x86"
    
    MACHINE_FEATURES = "kernel26 pci alsa usbhost"
    
    require conf/machine/include/tune-athlonmp.conf
    
    Written by Marcin Juszkiewicz on
  5. Paypal in Poland

    Paypal now officially supports Poland in both ways (previously it only allowed to send cash from Poland). So I decided to create account. From now I can accept donations via:

    I did not yet tested Paypal one.

    Written by Marcin Juszkiewicz on
  6. OEDEM 2006

    During “weekend” (Thursday — Sunday) I was in Berlin, Germany where OEDEM took place. It was nice to finally meet people whom I know via Internet for years.

    OEDEM 2006 people
    OEDEM 2006 people

    From left to right:

    • Richard `RP’ Purdie from OpenedHand
    • Tomas Frydrych from OpenedHand
    • I
    • Michael `Mickey’ Lauer
    • Koen Kooi
    • Florian Boor from Kernel concepts
    • Holger `Zecke’ Freyther

    We had many discussions about important OpenEmbedded things like problems, policies, creating some kind of OpenEmbedded Foundation for managing domains, trademarks, donations etc. Some of results are already happening:

    • removal of MAINTAINER fields (which we are going to replace with /MAINTAINERS file)
    • switching to task-base instead of the old one task-bootstrap to get clean images with just needed packages
    • removal of ipaq-pxa270 machines in favour of hx4700, hx2000 machines
    • new mailing lists got created:

    openembedded-issues with information from bugtracker — openembedded-users for all normal users which are not ready yet for openembedded-devel list — openembedded-announce for announcing

    More stuff will follow soon. I’m going to write documentation about task-base stuff, how to switch distro/machine to use it etc.

    Written by Marcin Juszkiewicz on
  7. Freeware definition of MS Windows world

    Today I looked for some tools for Microsoft Windows platform. Got some results marked as freeware so I looked deeply at them (why to pay if can get tool for free). And what did I saw? Definition of ‘freeware’ has been changed during last years… Now it means:

    Freeware: software which you can try for free but any usage need you to register and pay $$$.

    For me it is just ‘demo’, ‘trial’ version but not freeware.

    Written by Marcin Juszkiewicz on
  8. Home machine problems

    Some time ago my home machine started to be very unstable. I was getting random rootfs corruptions ('awk: file not found' on start for example), sometimes reboot during work…

    I treid to test all components:

    • S.M.A.R.T. tests to get info about HDD health — passed
    • backblocks test on /dev/hda1passed
    • memtest tests to get info how memory behave — passed
    • cpuburn tests gave 63.5°C on CPUpassed
    • cpuburn memory tests — FAILED

    After switching memory sticks in slots all memory tests were passed again. But system still kept failing ;(

    I’m lost of ideas how to fix it and more and more think about upgrading it to amd64 platform.

    Written by Marcin Juszkiewicz on
Page 92 / 105