Re: gcc-4.1: debian/patches/gcc-long-double.dpatch fails to apply

[Ludovic Brenta <ludovic@ludovic-brenta.org>]

Matthias Klose <doko@cs.tu-berlin.de> writes:
> sorry, I did check in changes on the sid branch, which are not target
> for etch. If we do want to make modifications for the etch target, we
> should work on branches/etch instead.

Well since Etch is frozen, the only thing that would justify an Etch
branch is a security vulnerability or other release-critical bug that
requires patching.  Currently I see no need for an Etch branch.

OTOH, the sid branch is supposed to work on sid, and apparently it
doesn't.  I can see at least three problems currently:

- gcc-long-double.dpatch does not apply cleanly
- build-depend on binutils (>= 2.17.50) which does not exist in sid
  (my earlier email to this list)
- On some architectures, build-depend on glibc (>= 2.4) which does not
  exist in sid (okay, there's 2.5 in experimental, but experimental is
  not sid, and no package should ever depend or build-depend on a
  package in experimental).

There may be other problems which I have not seen yet.

So, could you please explain exactly what the "sid" branch is supposed
to build on?  In other words, what does it assume?

I propose to revert the offending changes in the sid branch; if you
want to commit changes not intended for sid, create a new branch for
them.  Would you care to suggest a name for that new branch?

-- 
Ludovic Brenta.