| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
|
|
|
|
|
|
|
|
|
|
This moves all the limits handling to a dedicated class and the data to a
dedicated object, instead of embedding them into the scheduling code, where
they are hard to find and to edit.
|
|
suite name
|
|
architecture support
|
|
instead of db quering
|
|
notable changes:
* gen_packages_html() in common.sh got renamed to the singular version
gen_package_html() and now accept only one argument (a package name)
+ thus reproducible_blacklist.sh got fixed to pass one package at time
+ this greatly semplify the code, by removing *a lot* of crappy code
* _html_all_packages.py really becomed a 2-liner, with part of its login being
now either in _html_packages.py or removed.
|
|
Package class
This include some refactoring that change the interface, mainly:
* it does not care anymore about suites/architectures, it always builds for all
suites/architectures (it's only 0.122 for one package with no_clean=True)
+ this allows to remove *a lot* of crappy code
* it expects a list of Package instances, no more of strings (package names)
* check_package_status() got removed (as incorporated into the class)
This breaks quite everything, still need to update all the users of this function.
|
|
describing a package and its state
|
|
This reverts commit fc28ef4b067bd66daa7bd6f859c5aa28adaa9f57.
|
|
|
|
failed build to reschedule
|
|
|
|
|
|
on them
|
|
|
|
|
|
sid and experimental due to gcc5 transition + aptitude breakage"
This reverts commit 3364bd3fae718849a9cac881fd8d66db47794aae.
This does not yet return us to scheduling sid at the usual pace.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
building for
|
|
|
|
|
|
|
|
|
|
|
|
at all occurances of 'amd64' in the code anyway
|
|
|
|
|
|
others, to build them only once
|
|
|
|
for depwait and 404 packages
|
|
pages
|
|
|
|
|