Please accept this patch. 2.23 has some very annoying bugs related to source resolving (addr2line doesn't work at all, objdump -dS doesn't display sources etc...)
Thanks for your patch; however, it relates to an unadopted packaging system, which I will not use.
I have posted an experimental package set for binutils-2.24 here (sourceforge.net), but I have not yet added it to the mingw-get catalogue. Could you test (at least) the -bin package, for suitability and correct functionality, before I do that, please.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I just tested binutils-2.24, I locally installed exe and lang packages, and could successfully build the following packages: gnupg, pcre, openssl.
I also propose to make ranlib/ar default to deterministic builds.
This requires passing --enable-deterministic-archives to binutils/configure.
From configure --help
--enable-deterministic-archives
ar and ranlib default to -D behavior
Unfortunately I could not produce a patch as I did not found out which packaging the src package used.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I just tested binutils-2.24, I locally installed exe and lang packages, and could successfully build the following packages: gnupg, pcre, openssl.
Thanks for confirming this; I've now added that experimental package set to the mingw-get catalogue, as the current release.
I also propose to make ranlib/ar default to deterministic builds.
This requires passing --enable-deterministic-archives to binutils/configure.
Why? That would be a departure from our previous practice; furthermore, it would appear to be inconsistent with the configuration adopted by the major Linux distributors ... well, Debian anyway, but I don't recall this behaviour from my past use of Red Hat or SUSE either.
Unfortunately I could not produce a patch as I did not found out which packaging the src package used.
Any patch would need to be against the arch/mingw32/binutils-2.24-mingw32.pkgspec file, within the source tarball itself. However, I would need a very strongly defended justification from you, to even consider such a change ... I simply don't see, right now, why it would be beneficial to always mark all archive members as owned by root/root, with timestamps set to the beginning of the Unix epoch; if that seems useful to you, you may specify the "D" flag on invocation.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
First of all thanks for promoting the current binutils package to the public.
Concerning the deterministc builds. My intention is to be able to produce mingw binaries on a remote machine somewhere in the cloud. For that I have to have a possibility to verify that locally built binaries and binaries build in the cloud are identical.
And for that I'll need the --enable-deterministic-archives flag.
I can, and will, patch the binutils invocations so that the default value does not matter so much for me.
Thanks for your patch; however, it relates to an unadopted packaging system, which I will not use.
I have posted an experimental package set for binutils-2.24 here (sourceforge.net), but I have not yet added it to the mingw-get catalogue. Could you test (at least) the -bin package, for suitability and correct functionality, before I do that, please.
I just tested binutils-2.24, I locally installed exe and lang packages, and could successfully build the following packages: gnupg, pcre, openssl.
I also propose to make ranlib/ar default to deterministic builds.
This requires passing --enable-deterministic-archives to binutils/configure.
From configure --help
--enable-deterministic-archives
ar and ranlib default to -D behavior
Unfortunately I could not produce a patch as I did not found out which packaging the src package used.
Thanks for confirming this; I've now added that experimental package set to the mingw-get catalogue, as the current release.
Why? That would be a departure from our previous practice; furthermore, it would appear to be inconsistent with the configuration adopted by the major Linux distributors ... well, Debian anyway, but I don't recall this behaviour from my past use of Red Hat or SUSE either.
Any patch would need to be against the arch/mingw32/binutils-2.24-mingw32.pkgspec file, within the source tarball itself. However, I would need a very strongly defended justification from you, to even consider such a change ... I simply don't see, right now, why it would be beneficial to always mark all archive members as owned by root/root, with timestamps set to the beginning of the Unix epoch; if that seems useful to you, you may specify the "D" flag on invocation.
First of all thanks for promoting the current binutils package to the public.
Concerning the deterministc builds. My intention is to be able to produce mingw binaries on a remote machine somewhere in the cloud. For that I have to have a possibility to verify that locally built binaries and binaries build in the cloud are identical.
And for that I'll need the --enable-deterministic-archives flag.
I can, and will, patch the binutils invocations so that the default value does not matter so much for me.
See for example the discussion in http://comments.gmane.org/gmane.comp.gnu.binutils/63361.