[Build-common-hackers] Bug#508162: Bug#508162: Please document the "correct" order of rules and classes in debian/rules

[Jonas Smedegaard]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

clone 508162 -1
reassign -1 quilt
retitle -1 behavior of CDBS snippet differ based on placement
thanks

On Tue, Dec 09, 2008 at 09:11:52AM +0100, Fabian Greffrath wrote:
> Jonas Smedegaard schrieb:
>> In principle the order does not matter.
>>
>> But in reality, as you've experienced, it does.
>>
>> Please file individual bugreports for each such issue that you 
>> discover.
>
> OK, this is an individual bug report:
> Please make sure that patches are always applied *before* the build  
> system attempts to configure, build or clean a package.

No. Individual means file a separate bugreport for each concrete case of 
CDBS ordering changing actual behaviour without it being documented.

I have now cloned this general bugreport for the concrete case that you 
mentioned:

On Mon, Dec 08, 2008 at 15:10:58 +0100, Fabian Greffrath wrote:
>I need to include patchsys.quilt.mk *after* autotools.mk if one of my 
>patches affects to build system to run properly (e.g. to make 'make 
>distclean' work again). This behaviour is documented nowhere -- at 
>least I haven't found anything in the cdbs documentation.


I also filed that bugreport against quilt as Peter Eisentraut rightly 
point out that this specific case is tied to that other package.



>> At best such issues should be corrected to not exist, but if too 
>> impractical at least it should detect and emit a warning when loaded 
>> in a suboptimal order.
>
> For something to be able to detect a suboptimal order, the *optimal* 
> order should have been documented in the first place. ;)

Uhm, let me clarify: At best such issues should be fixed so that the 
principle of "order does not matter" is restored.

Only when order matters it should be documented how it affects the 
functionality of the snippet and perhaps that of other snippets.

I can even imaginge some cases of several orders being equally sane but 
for different purposes. Such cases should also be documented. Only if a 
non-sane ordering is detected it makes sense for a snippet to spew a 
warning.


Hope that helps. Both in the concrete case (the cloned bug) and for this 
kind of issues in general (bug#508162).


Kind regards,

   - Jonas

- -- 
* Jonas Smedegaard - idealist og Internet-arkitekt
* Tlf.: +45 40843136  Website: http://dr.jones.dk/

   [x] quote me freely  [ ] ask before reusing  [ ] keep private
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkk+gT4ACgkQn7DbMsAkQLi1dgCgjkmS7dZkFsGzK/A9/vE1wT2P
49AAnA9ehpHKlW8aTCqcna5SrZqotvMU
=4uiz
-----END PGP SIGNATURE-----