trunk: [-Wmismatched-tags] warnings

[su_v]

<quote>
A lot of the new [-Wmismatched-tags] warnings in the build log come from dodgy forward declarations where something that is really a class is declared as a struct or vice-versa. Strictly, it's not great to allow this, because class and struct imply different levels of default privacy (class="private by default"; struct="public by default"), so we shouldn't really get rid of the warning.

The good news is that these are easy fixes, and that a lot of it is duplicated. i.e., simply changing "class" for "struct" in one line of code in a widely-used header could fix thousands of these warnings.
</quote>
<https://bugs.launchpad.net/inkscape/+bug/992124/comments/33>

This report will track 'mismatched-tags' warnings for current trunk.

[su_v (suv-lp)]

Full build log (clean build).

[Alex Valavanis (valavanisalex)]

Note that there is an additional upstream issue in that was fixed in gtkmm_3.4 [1] but persists in the gtkmm 2.24 branch. I have asked for the fix to be backported [2].

[1] https://git.gnome.org/browse/gtkmm/commit/gtk/src/papersize.hg?id=0eec6fd533ef14f661fa143a57cf4b01d881be1f
[2] https://bugzilla.gnome.org/show_bug.cgi?id=709149

[Alex Valavanis (valavanisalex)]

Hopefully, all of our own remaining issues are fixed in lp:inkscape r12644.

[su_v (suv-lp)]

On 2013-10-01 02:41 +0200, Alex Valavanis wrote:
> Hopefully, all of our own remaining issues are fixed in lp:inkscape
> r12644.

Looks good to me: rev 12645 has only the two remaining 'mismatched-tags' warnings from gtkmm, now tracked upstream (thx for taking care of that - I had filtered them out intentionally, but didn't report them upstream).

Full build log from r12645 attached.

[su_v (suv-lp)]

Kris changed the bug status:
> Triaged → In Progress

Why? There is no more progress to be done in Inkscape. The remaining issue cannot be fixed here, it is an upstream issue. The commit is already in the git branch of gtkmm 2.24, and the report can be closed when (or if) a new gtkmm package is released.

[Kris (kris-degussem)]

Well ... actually there is not a real bug status on launchpad that imho describes the situation of this report. There were warnings that have taken care off (cfr. in progress), while the remaining issues are to be solved externally (triaged is not a good naming/category either). Just thought the bug status change better reflects that we do not need a change in our code. Feel free to reset the status ...

[su_v (suv-lp)]

Closing - fixed in trunk.

Upstream issue is in git, and can be applied by packagers if needed (a new release of gtkmm 2.24 seems unlikely to happen):
<https://git.gnome.org/browse/gtkmm/commit/?h=gtkmm-2-24&id=4fb3efa910a46364fa8ec4d60fcce557200e4dcb>

[su_v (suv-lp)]

Re-opening (new warnings with clang introduced in r13223).

[su_v (suv-lp)]

Build log r13229 attached (note: cast-align warnings from boost 1.55 with clang suppressed via #pragma in 'src/display/drawing-item.h').

$ grep '\[-W' 2014-03-29-werror_clang-13229.txt | grep mismatched-tags | wc -l
      53
$ grep '\[-W' 2014-03-29-werror_clang-13229.txt | grep mismatched-tags | grep SPKnot | wc -l
      47
$ grep '\[-W' 2014-03-29-werror_clang-13229.txt | grep mismatched-tags | grep PaperSizeTraits | wc -l
       6
$

[su_v (suv-lp)]

Proposed diff for r13229 (needs review).

[su_v (suv-lp)]

New warnings from r13223 fixed in r13230 by Alex Valavanis.

[su_v (suv-lp)]

Re-opening (new warnings with clang introduced in r13298):
$ grep '\[-W' 2014-04-24-mptrunk-quartz-clang-13299.txt | grep mismatched-tags | wc -l
     257
$ grep '\[-W' 2014-04-24-mptrunk-quartz-clang-13299.txt | grep mismatched-tags | grep SPStyle | wc -l
     253
$ grep '\[-W' 2014-04-24-mptrunk-quartz-clang-13299.txt | grep mismatched-tags | grep PaperSizeTraits | wc -l
       4

Proposed diff for r13299 attached (needs review).

[su_v (suv-lp)]

Warnings from SPStyle refactoring fixed in r13309. Thx @Tav for reviewing and committing!