Re: Taking patches in gcc-x.y to gcc-x.y-doc
On 10/20/2012 06:16 PM, Matthias Klose wrote:
> On 20.10.2012 08:56, Guo Yixuan wrote:
>> Hello Matthias and Debian GCC Maintainers,
>>
>> I'm preparing gcc's document packages for Debian, and have some questions
>> about patches. Because the documentation lives in a separate package from
>> gcc-x.y, I need to pick patches from gcc-x.y that are related to .texinfo
>> and some other files, and apply then in gcc-doc-x.y package.
>>
>> I've looked into gcc-4.[467]'s debian/ directory and know the patch/unpatch
>> process is handled by debian/rules.patch. It's quite complex and the actual
>> patch series differ between different archs. I want to know clearly which
>> patch to take, to keep gcc-x.y-doc accurate, so I seek confirmation/answer
>> to these doubts.
>>
>> 1. Does $(debian_patches) contains all the patches that are applied in a
>> specific build? I know $(hardening_patches) is only for Ubuntu, but what
>> are $(spu_patches) and $(no_spu_patches) for? 2. $(GFDL_INVARIANT_FREE) is
>> "yes", and most (or maybe all) document patches seem to be under condition
>> ifneq ($(GFDL_INVARIANT_FREE),yes) in rules.patch. 3. $(with_linaro_branch)
>> seems to be always empty for Debian. (But for gcc-4.4 on armhf, it's
>> "yes".) 4. As there're some arch-dependent doc patches, while gcc-x.y-doc
>> had better to be arch-indep, some patches need some modification, such as
>> this (changed by me, to make alpha-ieee-doc.diff suitable for arch-indep
>> document):
>
> run debian/rules patch and look at the generated series file.
The point is that I can't read just the generated series file, as the
document patches are not listed there, and series differ between archs.
To know what document patches to take, I have to read rules.patch in
addition. (Or should I set $(GFDL_INVARIANT_FREE) to "no" and
$(DEB_TARGET_ARCH) to every arch and run debian/rules patch many times?)
> spu patches are applied for powerpc only, so have a look at the doc patches
> if it makes sense to apply them.
Thanks, it seems that they are only for gcc-x.y-spu:powerpc, and not for
gcc-x.y:powerpc. Thus, I'm not going to apply spu document patches
(cell-branch-doc) in gcc-4.4-doc. (If anyone need that, it should live
in gcc-4.4-spu-doc, IMO.)
> we may use the Linaro branch for arm64 in gcc-4.7, but this is will be for
> jessie only.
Thanks for this info.
>> Index: gcc-doc/gcc/doc/invoke.texi
>> =================================================================== ---
>> gcc-doc.orig/gcc/doc/invoke.texi +++ gcc-doc/gcc/doc/invoke.texi @@
>> -11263,7 +11263,7 @@ values such as not-a-number and plus/minus infinity.
>> Other Alpha compilers call this option @option{-ieee_with_no_inexact}.
>>
>> -DEBIAN SPECIFIC: This option is on by default, unless +DEBIAN SPECIFIC:
>> (only for alpha architecture) This option is on by default, unless
>> @option{-ffinite-math-only} (which is part of the @option{-ffast-math} set)
>> is specified, because the software functions in the GNU libc math libraries
>> generate denormalized numbers, NaNs, and infs (all of which
>
> I'll update this.
Thanks for this, too.
Guo Yixuan
Reply to: