Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 231870 - new virtual/libstdc++ is necessary
Summary: new virtual/libstdc++ is necessary
Status: RESOLVED WONTFIX
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: New packages (show other bugs)
Hardware: All Linux
: High normal (vote)
Assignee: Gentoo Toolchain Maintainers
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-07-15 15:29 UTC by Doug Goldstein (RETIRED)
Modified: 2014-11-01 03:59 UTC (History)
7 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments
libstdc++.txt (libstdc++.txt,2.55 KB, text/plain)
2009-08-23 02:24 UTC, William Hubbs
Details
need testing. (libstdc++-v3-4.4.1.ebuild,2.03 KB, text/plain)
2009-10-03 19:41 UTC, Magnus Granberg
Details
New ebuild with GCC 4.4.2 and moved to slot 6 (libstdc++-v3-4.4.2.ebuild,2.45 KB, text/plain)
2009-10-18 12:18 UTC, Magnus Granberg
Details
Don't build crt*.o and libgcc with SSP (libstdc++-v3-Makefile.in.patch,3.97 KB, patch)
2009-10-18 12:19 UTC, Magnus Granberg
Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Doug Goldstein (RETIRED) gentoo-dev 2008-07-15 15:29:04 UTC
Since @system is removed from @world and we need to explicitly depend on libstdc++ in packages which link to it. Currently for gcc 3.4.x and higher libstdc++ there is no virtual to describe that depend. We'd want a virtual because in the future if gcc changes their C++ ABI we'll need to build the old versions.

So basically I propose we add a virtual/libstdc++ with a SLOT="4". In the future we can potentially move virtual/libstdc++v3 over to virtual/libstdc++ with SLOT="3" allowing for EAPI=1 slot based deps to be used. This is however in the future due to the fact that we don't want to bring EAPI=1 into system quiet yet.

This would be used only for packages that expressly link against libstdc++ (i.e. Qt)
Comment 1 Russell Harmon 2008-09-12 16:36:35 UTC
I agree that this is necessary.
Comment 2 Mark Loeser (RETIRED) gentoo-dev 2009-05-03 01:02:08 UTC
Before we can do this, we'll have to fix up some dependencies in the tree.  I'm almost positive not everything in the tree is depending on the exact version they need.

I'll file them as blockers against this bug when I get around to it, unless someone wants to do it before I do :)
Comment 3 Robin Johnson archtester Gentoo Infrastructure gentoo-dev Security 2009-08-22 18:31:51 UTC
As a workaround to avoid EAPI=1 for system at this time, how is this for a plan:

Phase1 (start of migration):
|| ( virtual/libstdc++v3 =virtual/libstdc++-3* )

Phase2 (ready to remove old v3 package):
=virtual/libstdc++-3*

Phase3 (later, when EAPI=1 slot deps enabled):
virtual/libstdc++:3
Comment 4 Robin Johnson archtester Gentoo Infrastructure gentoo-dev Security 2009-08-22 18:32:00 UTC
Forgot to add myself.
Comment 5 William Hubbs gentoo-dev 2009-08-23 02:24:52 UTC
Created attachment 201971 [details]
libstdc++.txt

This list of packages has been updated to depend on
~virtual/libstdc++-3.3.
Comment 6 Robin Johnson archtester Gentoo Infrastructure gentoo-dev Security 2009-08-26 23:31:38 UTC
halcy0n: so all the deps are fixed up now, what's next?

Probably adding the new slot I think?
Comment 7 Magnus Granberg gentoo-dev 2009-10-03 19:41:39 UTC
Created attachment 205958 [details]
need testing.

This ebuild pass compile.
Comment 8 Magnus Granberg gentoo-dev 2009-10-18 12:18:22 UTC
Created attachment 207467 [details]
New ebuild with GCC 4.4.2 and moved to slot 6
Comment 9 Magnus Granberg gentoo-dev 2009-10-18 12:19:33 UTC
Created attachment 207469 [details, diff]
Don't build crt*.o and libgcc with SSP
Comment 10 Anthony Basile gentoo-dev 2009-10-20 13:00:52 UTC
I've tested the resulting libraries from the ebuilds in the previous three comments against seamonkey-bin, opera, virtualbox-bin and bitdefender-console and found no issues.  I also tested them against my own C++ code and again found no issues whatsoever.
Comment 11 SpanKY gentoo-dev 2014-11-01 03:59:08 UTC
i think this is largely not necessary