Plasma 5.7.0 tars

Harald Sitter sitter at kde.org
Mon Jul 4 13:16:05 UTC 2016


On Mon, Jul 4, 2016 at 2:29 PM, David Edmundson
<david at davidedmundson.co.uk> wrote:
>
>
> On Mon, Jul 4, 2016 at 1:20 PM, Sebastian Kügler <sebas at kde.org> wrote:
>>
>> On Thursday, June 30, 2016 11:06:46 PM CEST Harald Sitter wrote:
>> > On Thu, Jun 30, 2016 at 4:45 PM, David Edmundson
>> >
>> > <david at davidedmundson.co.uk> wrote:
>> > > *sigh* seems so. Yet plasma-workspace is from the right branch and
>> > > it's
>> > > done by an automated script(!)
>> >
>> > Haven't we fixed that for beta already?
>>
> It was the cached kde_projects.xml that you fixed for the beta
> I hadn't pulled your changes as I needed some of my other local mods to get
> round some hardcoded paths. So partly my fault - though I still think it's
> very broken that we get version information from two different sources
> within the same script.

Yes. I mean no. I am confused.
If the projects.xml says that the stable branch of libkscreen is
Plasma/5.6 we can't simply override that on the release side of
things, l10n would still be generated from the wrong branch so you'd
misalign translations and source. That said, one could do a
consistency enforcement by failing if one of the things in a 'set'
(e.g. kde/workspace) is not in line with the rest.
BUT the plasma bash hell on top of releaseme bypasses most smarts of
releaseme todo with projects processing. In this particular case the
fact that releaseme can release a set (e.g kde/workspace) and handle
it as a 'set' is bypassed by the bash scripts wanting to decide what
to release and then calling tarme in a loop for each project.

So, assuming that is actually a smart way to do it (which it isn't
since it replicates information from projects.xml and allows for
additional human error and makes tarball creation slower), the bash
script would have to check consistency of the output coming out of
tarme (namely the release_data) and make sure that everything is using
the same branch.

HS


More information about the Plasma-devel mailing list