September 29, 2015

Gentoo's top-notch user community

One of the best parts of Gentoo for me is the community.

For example, I regularly receive email from people who volunteer to help with testing hard masked www-client/chromium ebuilds. FWIW you don't need to email me or the Gentoo Chromium team - just start testing and filing bugs. On the other hand, I do appreciate when people express interest in helping out.

Another example is helping with getting bugs resolved.

Bug #556812 looked rather mysterious - until Guillaume ZITTA found that "build" is a red herring, and in fact the "tracing" module being imported is a different one (from dev-python/tracing as opposed to chromium sources). It was an unfortunate names collision - once found, quite easy to fix.

In bug #553502 Johan Hovold found we need to require postproc USE flag for libvpx to avoid a crash.

See bug #551666 for some Alexey Dobriyan's gdb magic mostly based on a single line segfault report from dmesg...

These are just a few examples.

By the way, the area where we could use more help is arm architecture support. Some specific bugs where help is wanted are #555702#545368#520180, and #483130. Please report whether you can reproduce or not, post emerge --info from your system and the compressed build log in case build fails.

April 11, 2015

Tricks for resolving slot conflicts and blockers

Slot conflicts can be annoying. It's worse when an attempt to fix them leads to an even bigger mess. I hope this post helps you with some cases - and that portage will keep getting smarter about resolving them automatically.

March 5, 2015

More reliable handling of bash history across terminals and crashes

I've been occasionally hitting frustrating issues with bash history getting lost after a crash. Then I found this great blog post about keeping bash history in sync on disk and between multiple terminals.

tl;dr is to use "shopt -s histappend" and PROMPT_COMMAND="${PROMPT_COMMAND};history -a"

The first is usually default, and results in sane behavior when you have multiple bash sessions at the same time. Now the second one ("history -a") is really useful to flush the history to disk in case of crashes.

I'm happy to announce that both are now default in Gentoo! Please see bug #517342 for reference.

August 7, 2014

Can your distro compile Chromium?

Chromium is moving towards using C++11. Even more, it's going to require either gcc-4.8 or clang.

Distros like Ubuntu, Mageia, Fedora, openSUSE, Arch, CentOS, and Slackware are already using gcc-4.8 or later is their latest stable release.

On the other hand, Debian Wheezy (7.0) has gcc-4.7.2. Gentoo is using gcc-4.7.3 in stable.

I started a thread on gentoo-dev, gcc-4.8 may be needed in stable for www-client/chromium-38.x. There is a tracker for gcc-4.8 stabilization, bug #516152. There is also gcc-4.8 porting tracker, bug #461954.

Please consider testing gcc-4.8 on your stable Gentoo system, and file bugs for any package that fails to compile or needs to have a newer version stabilized to work with new gcc. I have recompiled all packages, the kernel, and GRUB without problems.

The title of this post is deliberately a bit similar to my earlier post Is your distro fast enough for Chromium? This browser project is pushing a lot towards shorter release cycles and latest software. I consider that a good thing. Now we just need to keep up with the updates, and any help is welcome.

July 19, 2014

Recovering from removed (and missing busybox)

I was experimenting in my arm chroot, and after a gcc upgrade and emerge --depclean --ask that removed the old gcc I got the following error:

# ls -l
ls: error while loading shared libraries: cannot open shared object file: No such file or directory

Fortunately the newer working gcc was present, so the steps to make things work again were:

# LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/usr/lib/gcc/armv7a-hardfloat-linux-gnueabi/4.8.2/" gcc-config -l
 * gcc-config: Active gcc profile is invalid!

 [1] armv7a-hardfloat-linux-gnueabi-4.8.2

# LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/usr/lib/gcc/armv7a-hardfloat-linux-gnueabi/4.8.2/" gcc-config 1 
 * Switching native-compiler to armv7a-hardfloat-linux-gnueabi-4.8.2 ...

Actually my first thought was using busybox. The unexpected breakage during a routine gcc upgrade made me do some research in case I can't rely on /bin/busybox being present and working.

I highly recommend the following links for further reading:

June 1, 2014

perl-cleaner slot conflict when upgrading perl (5.16 -> 5.18)

If you tried upgrading from stable amd64 to ~amd64 or otherwise done a big update of perl, you probably hit this weird perl-cleaner slot conflict:

# perl-cleaner --all
!!! Multiple package instances within a single package slot have been pulled
!!! into the dependency graph, resulting in a slot conflict:


  (dev-lang/perl-5.18.2:0/5.18::gentoo, installed) pulled in by
    =dev-lang/perl-5.18* required by (virtual/perl-IO-1.280.0:0/0::gentoo, ebuild scheduled for merge)
    ^              ^^^^^                                                                                                                                  
    dev-lang/perl:0/5.18=[-build(-)] required by (perl-core/version-0.990.800:0/0::gentoo, installed)
    (and 7 more with the same problems)

  (dev-lang/perl-5.16.3:0/5.16::gentoo, ebuild scheduled for merge) pulled in by
    =dev-lang/perl-5.16* required by (virtual/perl-Package-Constants-0.20.0-r3:0/0::gentoo, installed)
    ^              ^^^^^                                                                                                                                  
    (and 6 more with the same problem)

This is bug #506616, and the solution is to run the following command:

perl-cleaner --all -- --backtrack=30

January 20, 2014

System update report (last update a year ago)

I have some systems that I update more rarely than the others. Of course I highly recommend keeping up to date, especially for the security fixes. I've also written similar update reports in the past. You may want to read them and compare the experience:

Another 5-month update (December 2011)
Another report from rarely updated system (September 2012)