| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
| |
The previous regex would also match a non-specific one. Such doesn't
currently exist so this shouldn't currently make a difference.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
These were being used on rabbit.
|
|
|
|
| |
These were being used on duvel.
|
| |
|
|
|
|
| |
(need latest perl_checker)
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
This is now called and sadly hadn't yet been tested.
Just a minor typo and easy to fixup.
|
|
|
|
|
|
|
|
|
|
| |
the cmd line.
This allows you to mix moving backports and regular updates in one
command should you wish.
I don't expect this to be used much, but it's more correct to do it
this way I think.
|
|
|
|
|
| |
As --no-confirm may be used from automated scripts, better bail out here
rather than waiting for user input that will never come.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This allows us to specify multiple packages (from various distro
releases and sections) to be processed.
All the arguments are parsed into arrays and then these arrays are
looped over and processed all at once.
If the --sync option is given then it will only update the affected
hdlists and md5sums once.
The idea next is to write a wrapper in python which will take advisory
bug numbers as arguments and then:
1. Ensure each has an ID assigned (this has to be done first)
2. Calculate all the packages involved.
3. Run this script with the --sync --no-confirm options
4. Close all the bugs with a comment + link to the Advisory (hence the
need for the advisory ID in step 1 and the desire to use python as
this is how the git->bz integration works).
This scheme should save a lot of time, but stops just short of fully
automating things.
Sadly the cronjob that runs "mgaadv mksite" would likely have to be
disabled, but we should be able to add a new status field that does
these moves too and incorporate that into mgaadv and thus provide full
automation \o/
|
| |
|
|
|
|
|
| |
Also some cosmetic changes and fixes for the handling of the sanity check
response.
|
|
|
|
|
|
|
|
| |
This is needed specifically in the case of update_hdlist() where a previous run
would set the ofdeps variable which would then be reused on the next run.
This is typically not a big problem but it did mean that standard media file-deps
file was used when updating debug media which seems incorrect.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
The -update suffix was no longer relevant since support for other subsections
than 'updates' was added.
Also install it :)
|
|
|
|
| |
All the scripts on the build server behaved this way so we may as well too.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This replaces two scripts found on the build system: mga-move-backports-{nosync,sync-mail}
The latter scripts name includes the word 'sync' but it didn't actually do the sync.
but as this script supports a --no-sync option, this is now fully optional.
There seemed to be a stray $ in the old file. I don't think it was deliberate.
The mail sent to the backports ML is nicer than before as it doesn't include
output from raw commands, just a human readable equiv.
This also still sends a mail for audit purposes regardless of the nice
announce email
|
|
|
|
| |
Most of the code is the same/shared so it makes sense to do this in one script
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This tidies up the script and modernises it. It adds various options that allow more
practical usage. This includes a specific --dry-run option to just test things
and a --no-sync option which allows the usage we expect when processing updates
(the sync process takes a long time and thus it's better to leave it until
several updates are processed).
It also adds support for a slightly nicer email notification and adds an option
(--no-confirm) to avoid the confirmation step if you pass an exact srpm name
into the script rather than just the %{name} part of it which will be useful
when this process is automated.
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
These are the same scripts that were present on the live server and have
thus proven useful.
That said, common code has been factored out and centralised and the syntax
has been neatened up.
|
|
|
|
|
| |
This keeps the config file but ensures it's not 100% needed via
default values
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
doing less work in the regexp engine and less path walks
elapsed time consistently go down from 0.70 to 0.62 on scores of run
(mainly a win in time spend in the kernel)
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
collected so don't return anything...
|
| |
|
| |
|