[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Automatic Debug Packages Creation and Handling



Josselin Mouette wrote:
> Le dimanche 19 juillet 2009 à 13:17 +0200, Martin Pitt a écrit :
>> Hello,
>>
>> Emilio Pozuelo Monfort [2009-07-03  0:25 +0200]:
>>> In order to automatize the process, we'll modify the helper tools (debhelper and
>>> CDBS) 
>> I don't think you should rely on CDBS for that. In Ubuntu's
>> implementation, we currently divert dh_strip, so we just rely on
>> debhelper.
> 
> Since cdbs always uses dh_strip -pfoo, it is not possible to build one
> debug package per source package without changes in cdbs.

Yeah. Those should be minimal (if at all needed) though.

>>> - There's the question as to whether build one .ddeb per source package, or one
>>>   per binary package.
>> I vote for the latter, for two reasons:
>>
>>  * Users think in terms of binary packages.
> 
> Since we want to make their use completely transparent in the end, that
> doesn’t change many things for users.

Also it would be possible to extend apt-get/aptitude with a debug option so that
e.g. "apt-get debug vinagre" installs the debugging symbols for vinagre, without
having to think of source packages.

>>  * Often, source packages multibuild the same thing with different
>>    options (think of the mplayer-k6 vs. mplayer-686 packages in the
>>    past, or exim4-daemon-light vs. -heavy). You can't install both of
>>    the debug symbols at the same time.
> 
> You can install both at once, since they will have different build IDs.

We have a problem with other debug information, like python _d.so extensions. So
we need to think how we can solve issues like that, or use one ddeb per binary
package (the biggest drawback for this is the number of ddebs and bloating a
changes file. The former doesn't sound too bad (the size won't grow), the latter
could be 'solved' by removing one or two of the three hashes we currently have
in them?)

Cheers,
Emilio

Attachment: signature.asc
Description: OpenPGP digital signature


Reply to: