On Sat, 10 Oct 2009 10:54:18 -0700
Ned Ludd <email@example.com> wrote:
> Luca's and Zac's comments work for me.
> Either PMS seems to be about documenting ebuild syntax. If we force
> in a change for mtimes then it's no different than forcing a given
> syntax for VDB/binpkg handling etc. And I have a feeling we don't
> really want to open that can of worms.
Uhm. Two things.
First: this is not about existing syntax. Different Portage versions do
different things with mtimes, so currently ebuilds can't rely upon any
particular behaviour. The proposal is about standardising behaviour for
EAPI 3, which would allow EAPI 3 ebuilds to rely upon mtimes being
handled in a particular way. The previous Council rejected this
proposal, but Ulrich wants it reconsidered. If it is reconsidered, the
question is what behaviour we want to standardise. Going with "what
Portage does" is undesirable for two reasons -- first, it leads to
files with timestamps like 1 Jan 1970 being merged to /, and second,
it's undefined behaviour for any file that's modified by the package
manager (e.g. for stripping, fixing WORKDIR mentions, compressing docs,
repairing QA violations and so on -- there are currently no restrictions
on what a package manager can tidy up).
Second, VDB and binary packages have nothing to do with PMS.
If you just want to document "what Portage does", then PMS and EAPI 3
need no changes, since "what Portage does" depends upon what Portage
version you're using.