[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Autobuilder update



Hi,

Michael Banck, le Sat 03 Feb 2007 11:07:30 +0100, a écrit :
> This means we are now able to research in detail which issues are
> responsible for how many package build failures, and which as-of-yet
> unported packages are responsible for blocking how many other packages
> due to being a Build-Dependency for them.
> 
> The latter has been very nicely visualized by Samuel Thibault here:
> http://dept-info.labri.fr/~thibault/tmp/graph-radial.ps

Just a note here: not all dependencies are represented here (else it
would be just a _total_ mess :) ). What is represented exactly is only:

- the dependencies shown on the Dep-Wait buildd page
- the dependencies needed by Building packages

so it's far from complete (in particular the sometimes long dependency
chains) but it already gives a good statistical idea. About the colors:

- red: Failed (probably needs to be patched)
- white: Dep-Wait
- black: Not-For-Us (so dependencies on this package should be dropped)
- orange: Building (hence undecided state)
- green: Uploaded (so _will_ be available)
- magenta: doesn't exist ! (so the dependency is broken, probably just
because the dependency just needs to be fixed in the Dep-Wait page. For
instance, libicu34 is missing because it's now called libicu36,
and since it will probably be called libicu37 sooner or later...)

> An overview of all (currently 1269) build failure reasons can be found
> here: http://unstable.buildd.net/index-hurd-i386.html

Just a short note here: before jumping on looking at the build log,
please check that nobody did the job yet :)

Check at least the alioth database:
http://alioth.debian.org/tracker/?atid=410472&group_id=30628&func=browse
http://alioth.debian.org/tracker/?atid=411594&group_id=30628&func=browse

and the bug report database:
http://bugs.debian.org/src:name_of_the_source_package

(so you can see that part of the biggest blockers already have a patch)

> I will make an analysis on this data in the next days to give us some
> more information on what failures we most frequently run into.

In the end, I would like to thank Michael for his incredible work
keeping fixing buildd and requeuing packages :)

Samuel



Reply to: