[Piuparts-devel] remaining updates targeting for jessie

[Andreas Beckmann]

On 2014-11-30 23:14, Holger Levsen wrote:
> Hi Andreas,
> 
> thanks for all your patches! :-) despite getting grmbling below, I'm very 
> happy to have received them and look forward to merge/chery-pick them 
> tomorrow..
> 
> On Sonntag, 30. November 2014, Andreas Beckmann wrote:
>> On 2014-11-29 12:02, Holger Levsen wrote:
>>> anything else which should go into 0.61 now? I'll probably wait for the
>>> upload til monday until you tell me earlier you have no pending
>>> changes...
>>
>> Andreas Beckmann (13):
>>       detect_piuparts_issues: look for a new tar error message
not sure is this is another disk full issue or some other problem
>>       use mysql for otrs2 installation in squeeze
your workaround broke otrs2 testing in squeeze - let's limit the
unsupported (aka failing) bits to a few upgrade paths
>>       p: preserve domain setting in /etc/resolv.conf
bugfix
>>       add purge workaround for polgen-doc (lenny, non-free)
for the purpose of p.d.o ducomentation only
>>       scripts-leftovers: enable database purging for more mysql servers
preseeding must change with package name change
>>       scripts: update special cases for experimental
again
>>       Revert "custom-scripts: add debugging output for resolvconf. (See
>> #660350)"
>>       scripts: remove debugging output for resolvconf
or do we still need these debugging bits?
>>       distros.conf: add *updates, *backports, *proposed for jessie
we really need them
>>       p.conf.anbe: update various limits
documentaions only, primarily stop squeeze rescheduling, but other
changes have accumulated
> 
> why? please explain these changes...
> 
> I'm not sure this is suitable/very relevant for 0.61..
> 
> 
>>       add changelog entry for scripts updates
trivial
>>       add trailing colons to bug template subjects to encourage giving
>> more details
> 
> I rather prefer colons in the beginning, "$pkg: fails to install" reads much 
> better than "fails to install: $pkg"

I use reportbug to file bugs, so I get '^$pkg: ' for free :-)
But I find 'foo: fails to install' rather useless if I later have too
look at bugreports again. I'd prefer something like

#767826:
python3-django-assets: fails to install: SyntaxError: Missing
parentheses in call to 'print'

Especially for high-volume reports: fails to install, leftover crap
after purge, ...

I usually add the colon manually and add more information to the
subject, but having it at the end of the pasted string would actually
remind me to do this

> </nitpicking>
> 
> I'm also not sure this is suitable/relevant for 0.61.

That's mostly "documentation updates"

>>       add fails_to_upgrade_from_jessie* bug templates
> 
> not yet merged, because I've just said on #debian-release
> 
> [23:06] <      h01ger> | so andreas beckmann did some 13 useful commits to 
> piuparts (12 going into the scripts running on piu.d.o) and then he 
>           made another one, adding a colon to 16 bug-templates
> [23:07] <      h01ger> | i'm not happy as i think its unneeded and annoying to 
> you
> [23:07] <      h01ger> | but will still take it as its the least pain and 
> because the commit is probably useful and easy to filterdiff...
> [23:08] <      h01ger> | you have some minutes to scream ;)
> 
> 
>> Thereafter we have one more patch for jessie that should be the last one
>> to be cherry-picked before you s/UNRELEASED/unstable/:
>>
>> pick-for-jessie
>> Andreas Beckmann (1):
>>       distros.conf: switch 'testing' to 'stretch', deactivate it for stable
...
> I'm not convinced this is the proper route to address the underlying issue. 
> Not sure how to...

The *only* benefit of that alias is to reuse jessie_$arch.tar.gz for
testing_$arch.tar.gz. The only place where this is used is the
testing2sid test. And it needs piuparts-slave.

for piuparts itself it does not matter if you use "-d testing" or
"-d jessie", both are available from the mirrors

>> I hope that's it for 0.61 and jessie.
>> (If we need to change anything w.r.t eatmydata, we could branch this off
>> from 0.61 in case develop already contains non-freeze-qualifying commits.

looks like eatmydata got a pre-approval, so this should be good

> basically they are all fine...
> 
> Maybe I'm just lazy and should just cherry-pick those 13 "good" commits and 
> simply refuse this one apple which ain't ripe yet ;-)

please give the resolv.conf bits a bit testing before uploading - it
looks good on my side now

the only thing left to do is looking at debsums for distupgrades of the
apache stack that renames conffiles and moves them to a different
package at the same time - and I would prefer to not use --warn


Andreas