Opened 8 years ago

Closed 8 years ago

Last modified 8 years ago

#49695 closed defect (wontfix)

akonadi @1.13.1.20141210_1: build failure on ML

Reported by: mkae (Marko Käning) Owned by: NicosPavlov
Priority: Normal Milestone:
Component: ports Version:
Keywords: Cc: RJVB (René Bertin)
Port: akonadi

Description

The ML buildbot failed on akonadi:

Error: akonadi does not support your selected MacPorts C++ runtime. libc++ must be selected and C++-based ports built against it.

due to the changes made in r142082.

Change History (7)

comment:1 Changed 8 years ago by RJVB (René Bertin)

should those "r142082" style links point to a relevant location?

comment:2 Changed 8 years ago by NicosPavlov

Resolution: wontfix
Status: newclosed

This is the intended behaviour, as defined in the cxx11 PortGroup, which has been included as a conclusion of #49567 (I forgot to include the ticket in the commit log). Akonadi requires cxx11, which cannot be built with libstdc++.

comment:3 Changed 8 years ago by RJVB (René Bertin)

Are you sure that is not only for the tests? I cannot remember what was the latest version of akonadi that I built on 10.6, but I do remember that I built quite a few KDE ports with configure.compiler=macports-gcc-4.7 (or higher). Despite Ryan's misgivings in #49567 that never caused me any issues.

If it's not only the tests, a way out on OS X 10.6 would be to restrict the port version to the latest akonadi version that doesn't require libc++ , at least via a variant .

comment:4 Changed 8 years ago by NicosPavlov

Yes, sure. Akonadi fully requires c++11.

comment:5 Changed 8 years ago by RJVB (René Bertin)

A quick check of the svn log shows that r126425 is probably the latest version that ought to build on 10.6 .

We could clamp the port at the corresponding akonadi version on 10.6 (and depending on the port myself I'm inclined to say we probably should).

Another question/remark, related only because I saw nothing relevant in the svn log : I just applied an over-due update to port:boost, and could no longer start akonadi afterwards. All akonadictl did was print a backtrace, which fortunately led me to try a rebuild, which fixed the issue. Turns out it links to libboost_program_options, but rev-upgrade didn't catch any issues.

comment:6 Changed 8 years ago by NicosPavlov

I don't think r126425 specifically would change anything to the current situation. The tests were then deactivated anyway as they were causing other issues (see r127727). Assuming that version 1.13.0 of akonadi could work without c++11 support apart from the tests (which would have to be confirmed), then the build for older systems got most likely broken in r134234 when upgrading to a more recent git commit.

While the port does not build on the buildbot, it does not mean it cannot be built on 10.6, as explained in the cxx11 PortGroup (see https://trac.macports.org/wiki/LibcxxOnOlderSystems).

comment:7 Changed 8 years ago by RJVB (René Bertin)

You're probably right about that last part; as far as I can see, my 10.6 VM has the following versions installed:

akonadi-1.13.0_0+docs+mariadb55.darwin_10.x86_64.tbz2 
akonadi-1.13.1_20141023+docs+mariadb55.darwin_10.x86_64.tbz2 
akonadi-1.13.1_20141111+docs+mariadb55+quiet.darwin_10.x86_64.tbz2 

So I even managed to build an 1.13.1 version - presumable with a gcc >= 4.7 port .

I haven't been installed binary packages for my main KDE4 ports since a long time ...

Note: See TracTickets for help on using tickets.