[gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Andreas K. Huettel]

[-- Attachment #1: Type: text/plain, Size: 1587 bytes --]

So here's the agenda. It's been mostly assembled by tamiko, who may 
unfortunately not make it tonight. I'll try to stand in...

 - a short hello :-)

 - Outstanding CVEs for binutils, elfutils, gcc, and glibc [1]
   * Who handled CVEs in the past?
   * Who wants to take care of individual packages?
   * Stabilizing newer binutils and glibc is urgent.

 - current state of gcc porting (e.g. troubles with +ssp on alpha)
   * We have to fix the gcc compilation issues
   * Open gcc bugs [2]

 - full multilib compliant toolchain.eclass for gcc-7

 - How to tackle all open toolchain bugs? [3]
   * Volunteers for individual packages? (e.g. binutils, glibc, etc.)

 - Globally setting -std=c++14 in upcoming 17.0 profiles

 - What about quick (bi-)monthly meetings (~15 minutes) on IRC to keep 
ourselves updated?


Cheers, 
Andreas
  


[1] https://bugs.gentoo.org/buglist.cgi?email1=toolchain
%40gentoo.org&email2=security
%40gentoo.org&emailassigned_to2=1&emailcc1=1&emailtype1=substring&emailtype2=substring&list_id=3538312&query_format=advanced&resolution=---

[2] https://bugs.gentoo.org/buglist.cgi?email1=toolchain
%40gentoo.org&emailassigned_to1=1&emailtype1=substring&list_id=3538318&query_format=advanced&resolution=---
&short_desc=gcc&short_desc_type=allwordssubstr

[3] https://bugs.gentoo.org/buglist.cgi?email1=toolchain
%40gentoo.org&emailassigned_to1=1&emailtype1=substring&list_id=3538326&query_format=advanced&resolution=---


-- 
Andreas K. Hüttel
dilfridge@gentoo.org
Gentoo Linux developer (council, perl, libreoffice)

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 981 bytes --]

Re: [gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Alexis Ballier]

On Sat, 20 May 2017 19:38:01 +0200
"Andreas K. Huettel" <dilfridge@gentoo.org> wrote:
>  - Globally setting -std=c++14 in upcoming 17.0 profiles

A bit late to the party, but what was the outcome of the meeting, esp.
this part ?

I'm not sure if default CXXFLAGS would make such a difference since
most people would override them. OTOH, masking <gcc-6 (*) on those
profiles makes a lot of sense: it'll default to gnu++14 in a better way
than with profiles and since I think one of the points of the new
profiles is to force default-pie, it'll also avoid people using gcc5-
that would not build pie by default and have issues when switching to
gcc6+



(*) not sure if masking is the proper way, but I can't think of
anything better atm

Re: [gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Matthias Maier]

On Fri, May 26, 2017, at 13:07 CDT, Alexis Ballier <aballier@gentoo.org> wrote:

> On Sat, 20 May 2017 19:38:01 +0200
> "Andreas K. Huettel" <dilfridge@gentoo.org> wrote:
>>  - Globally setting -std=c++14 in upcoming 17.0 profiles
>
> A bit late to the party, but what was the outcome of the meeting, esp.
> this part ?
>
> I'm not sure if default CXXFLAGS would make such a difference since
> most people would override them.

Exactly. Setting this in CXXFLAGS is not a great idea. Forcing via SPEC
file is equally hacky.


> OTOH, masking <gcc-6 (*) on those profiles makes a lot of sense: it'll
> default to gnu++14 in a better way than with profiles and since I
> think one of the points of the new profiles is to force default-pie,
> it'll also avoid people using gcc5- that would not build pie by
> default and have issues when switching to a gcc6+

This is exactly what is planned :-)

Best,
Matthias

Re: [gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Kristian Fiskerstrand]

[-- Attachment #1.1: Type: text/plain, Size: 391 bytes --]

On 05/26/2017 08:07 PM, Alexis Ballier wrote:
> A bit late to the party, but what was the outcome of the meeting, esp.
> this part ?

Unofficial log from meeting:
https://download.sumptuouscapital.com/gentoo/tmp/gentoo-toolchain.log.txt

-- 
Kristian Fiskerstrand
OpenPGP keyblock reachable at hkp://pool.sks-keyservers.net
fpr:94CB AFDD 3034 5109 5618 35AA 0B7F 8B60 E3ED FAE3


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Walter Dnes]

On Fri, May 26, 2017 at 08:29:59PM +0200, Kristian Fiskerstrand wrote
> On 05/26/2017 08:07 PM, Alexis Ballier wrote:
> > A bit late to the party, but what was the outcome of the meeting, esp.
> > this part ?
> 
> Unofficial log from meeting:
> https://download.sumptuouscapital.com/gentoo/tmp/gentoo-toolchain.log.txt

  The irc log mentions problems with some ebuilds under some versions of
GCC.  In a similar vein, are there any anticipated issues with binaries
built under older GCC versions running on a system that is otherwise
all-GCC-6.3.0?  I'm thinking of pre-built binaries like Pale Moon (from
their website), firefox-bin, seamonkey-bin, openoffice-bin,
libreoffice-bin, libreproject-bin, etc.

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

Re: [gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Andreas K. Huettel]

[-- Attachment #1: Type: text/plain, Size: 1815 bytes --]

Am Montag, 29. Mai 2017, 19:32:48 CEST schrieb Walter Dnes:

> 
>   The irc log mentions problems with some ebuilds under some versions of
> GCC.  In a similar vein, are there any anticipated issues with binaries
> built under older GCC versions running on a system that is otherwise
> all-GCC-6.3.0?  I'm thinking of pre-built binaries like Pale Moon (from
> their website), firefox-bin, seamonkey-bin, openoffice-bin,
> libreoffice-bin, libreproject-bin, etc.

Yes, but you already should have these issues (and it's a different problem).

A] Here, we've been talking about changes in the default language standard. 
-- GCC up to 5.* defaults to -std=gnu98, while GCC 6 defaults to -std=gnu14
-- ICU 59 >requires< its reverse dependencies to build with at least -std=c+
+11
-- So all packages that a) use gcc-4 or gcc-5, and b) do not in the ebuild 
"manually" add something like -std=c++11 or -std=c++14 or -std=gnu14 will fail 
to *build*.

B] What you are talking about is a change in the ABI, the binary interface of 
the libraries. That happened as well (when gcc-5.4 was stabilized), and was 
the reason for the "rebuild your stuff" newsitem.
-- GCC-5 and later internally has a different representation of C++ stuff, 
compared with gcc-4
-- Building a library with gcc-4, and an executable that links to it with 
gcc-5, or vice versa, may fail with "missing symbols".
-- If you have a binary app built with gcc-4, and your libraries e.g. are 
built with gcc-5, you may run into the same problem (and your app won't 
start). 
-- This is why e.g. unbundling of libraries in vmware-workstation was at some 
point not possible anymore.
-- I don't know of any workaround.

HTH

-- 
Andreas K. Hüttel
dilfridge@gentoo.org
Gentoo Linux developer (council, perl, libreoffice)

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 981 bytes --]

Re: [gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Kristian Fiskerstrand]

[-- Attachment #1.1: Type: text/plain, Size: 737 bytes --]

On 05/30/2017 12:05 AM, Andreas K. Huettel wrote:
> -- So all packages that a) use gcc-4 or gcc-5, and b) do not in the ebuild 
> "manually" add something like -std=c++11 or -std=c++14 or -std=gnu14 will fail 
> to *build*.

This isn't really different from [Qt 5.7 requirements] and is
fundamentally an upstream bug if not checked for during configure and
automake using e.g [ax_cxx_compile_stdcxx_11].

References:
[Qt 5.7 requirements]
https://bugs.gentoo.org/589412

[ax_cxx_compile_stdcxx_11]
https://www.gnu.org/software/autoconf-archive/ax_cxx_compile_stdcxx_11.html

-- 
Kristian Fiskerstrand
OpenPGP keyblock reachable at hkp://pool.sks-keyservers.net
fpr:94CB AFDD 3034 5109 5618 35AA 0B7F 8B60 E3ED FAE3


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [gentoo-dev] toolchain meeting agenda for today 19:00 UTC #gentoo-toolchain

[Andreas K. Huettel]

[-- Attachment #1: Type: text/plain, Size: 1091 bytes --]

On May 30, 2017 10:09:14 AM GMT+02:00, Kristian Fiskerstrand <k_f@gentoo.org> wrote:
>On 05/30/2017 12:05 AM, Andreas K. Huettel wrote:
>> -- So all packages that a) use gcc-4 or gcc-5, and b) do not in the
>ebuild 
>> "manually" add something like -std=c++11 or -std=c++14 or -std=gnu14
>will fail 
>> to *build*.
>
>This isn't really different from [Qt 5.7 requirements] and is
>fundamentally an upstream bug if not checked for during configure and
>automake using e.g [ax_cxx_compile_stdcxx_11].
>
>References:
>[Qt 5.7 requirements]
>https://bugs.gentoo.org/589412
>
>[ax_cxx_compile_stdcxx_11]
>https://www.gnu.org/software/autoconf-archive/ax_cxx_compile_stdcxx_11.html
>
>-- 
>Kristian Fiskerstrand
>OpenPGP keyblock reachable at hkp://pool.sks-keyservers.net
>fpr:94CB AFDD 3034 5109 5618 35AA 0B7F 8B60 E3ED FAE3

Well, even if it were "fixed" in all upstream packages, we'd still have to sprinkle -std=c++11 over all ebuilds. I'd rather do the inverse for things not future proof...
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

[-- Attachment #2: Type: text/html, Size: 1538 bytes --]