Anssi Hannula skrev 6.12.2011 01:29:
On 06.12.2011 00:56, Thomas Backlund wrote:
Now,
here comes the question about backports once again.
We are now 6+ months into Mageia 1, and we are nowhere closer to opening
backports that we were at Mageia 1 release time.
Because of that there are 3rdparty repos popping up everywhere...,
something we hoped to avoid atleast partly when starting this project.
And at current rate we will probably release Mageia "infinity"
before backports is opened.
It has been delayed because of needed infrastructure changes,
something no-one have had time to do so far...
I know there is "only some coding missing" and "someone should
do it", buth truthfully there are only a few that knows the
code used in the buildsystem enough to actually make it happend,
and they are already othervise busy or overloaded...
(this is no rant against them, as all here are using their
personal free time to help out)
The biggest problem for me is that it is really difficult to test any
changes, one'd need a mockup of the whole buildsystem... and I don't
want to propose any patches blind :)
And to be honest I dont see that changing anytime soon...
So here is a suggestion to get some value to our endusers:
we add a backports branch on svn
So packages for backports would use:
svn.mageia.org/packages/backports/1/<package>/{current,pristine,releases}
and allow that to be used for backports.
Using a separate branch is also a cleaner way of providing
backports, and makes it easy to separate changes needed only
for Cauldron (or backports).
Hm, how does this help with enabling backports (i.e. compared to simply
using cauldron repo)?
It was suggested to prevent direct submissions from cauldron as
current BS is not capable of blocking cauldron -> updates_testing
if we add cauldron to supported url
So a separate backports branch would be created to "work around" this.
(yes I know you could do svn cp and then submit, but atleast it would
prevent submissions done by mistake)
And if we want to extend the checks, maybe mgarepo could be updated to
only allow submission to backports_testing from backports branch until
server side checking is in place. (the same "connection" could probably
be done between updates_testing and updates too)
Anyway, would the above branch work like this:
e.g. lets imagine I want to, in order:
1. provide foobar 1.1 to backports.
2. cauldron foobar upgraded to 1.2, some patches added, some removed,
and other changes
3. provide foobar 1.2 to backports
4. 1.3 to cauldron and backports
To backport foobar 1.1, I guess I'd simply do
svn copy svn+ssh://svn.mageia.org/svn/packages/cauldron/foobar \
svn+ssh://svn.mageia.org/svn/packages/backports/1/foobar
and then the submit command.
When time comes to provide foobar 1.2 to backports, the options are:
Option 1:
svn del svn+ssh://svn.mageia.org/svn/packages/backports/1/foobar
svn copy svn+ssh://svn.mageia.org/svn/packages/cauldron/foobar \
svn+ssh://svn.mageia.org/svn/packages/backports/1/foobar
and submit. Changelog is the same as in cauldron.
Option 2:
svn checkout \
svn+ssh://svn.mageia.org/svn/packages/backports/1/foobar/current \
foobar
cd foobar&& svn merge -r prevrev:HEAD \
svn+ssh://svn.mageia.org/svn/packages/cauldron/foobar/current
svn commit
# copy all the log entries from the cauldron interval to the log msg
and submit.
Note: results in broken changelogs unless one does manual
markreleases, but changelogs are already broken both in mga1 updates
and in cauldron (youri issue [1]). This is more notable here, though,
as backports happen generally more often multiple times than updates.
Repeat with 1.3.
Correct?
I guess this is up to the maintainer how he/she wants to maintain the
backports of his package as it can also be Option 3: maintain backports
branch separately (depending on package).
For example, for me I'd like to provide kernel-rt-3.0.x as backports,
but would probably not provide 3.2.x at all for Mageia 1.
(sidenote, kernel-rt is a little tricky here as it was available as
2.6.33 series in 2010.2, so it could go to updates. but since it
would also need backported nvidia/fglrx (if they work that is),
and the /proc/net/route info changed in 2.6.39 so it would "break"
network detection too, and version 3.* could break some scripts
checking for 2.*, so I'd rather submit it as a backport.)
[1] https://bugs.mageia.org/show_bug.cgi?id=2633
Yeah, this is another thing that needs to be fixed regardless of
what we do with backports. for now we have just disabled markrel
for updates so it does not mess up the changelogs even more.
--
Thomas
