Bug 430194 – games-engines/qtads-2.1.4: version bump

• Product: Gentoo Linux
• Status: RESOLVED FIXED

[Description Nikos Chantziaras 2012-08-06 15:38:23 UTC]

Created attachment 320570 [details, diff]
Patch against qtads-2.1.3.ebuild

QTads 2.1.4 was released on 2012-08-06.

I always avoid doing 0-day bump requests, but an ebuild change is needed that needs to be documented :-)

The current ebuild works fine, but won't install the newly provided file icons and MIME types.  I've added an ebuild to the interactive-fiction overlay:

http://repo.or.cz/w/gentoo-interactive-fiction.git/blob_plain/HEAD:/games-engines/qtads/qtads-2.1.4.ebuild


I'm also attaching a diff against the current 2.1.3 ebuild.

[Comment 1 Julian Ospald 2012-08-06 17:08:00 UTC]

fdo-mime functions calls are missing in pkg_postrm

use a proper doicon call: newicon -s 256 qtads_256x256.png ${PN}.png (will install into /usr/share/icons/hicolor/256y256/apps)
unless the doins -r line already installs there
(either way you need to update gnome2 icon cache via gnome2-utils.eclass, see games-arcade/mari0 ebuild for example)

doins is missing "|| die" (or use EAPI=4)

[Comment 2 Julian Ospald 2012-08-06 17:10:40 UTC]

the term "gnome2 icon cache" was misleading, I mean gtk icon cache which is mandatory when installing into /usr/share/icons/... because icons might otherwise not show up. Programs look up the icon cache instead of searching the subdirs (except we install into /usr/share/pixmaps but /usr/share/icons is more modern)

[Comment 3 Nikos Chantziaras 2012-08-06 17:57:18 UTC]

Thanks for the pointers, Julian!  I've updated the ebuild (and bumped EAPI to 4.)

I suppose there's nothing that can be done for the KDE icon cache?  It doesn't invalidate until I manually:

  $ rm /var/tmp/kdecache-$USER/icon-cache.kcache

and restart KDE.

[Comment 4 Nikos Chantziaras 2012-08-06 17:58:08 UTC]

Created attachment 320574 [details, diff]
Patch against qtads-2.1.3.ebuild

[Comment 5 Julian Ospald 2012-08-06 19:08:11 UTC]

(In reply to comment #3)
> Thanks for the pointers, Julian!  I've updated the ebuild (and bumped EAPI
> to 4.)
> 
> I suppose there's nothing that can be done for the KDE icon cache?  It
> doesn't invalidate until I manually:
> 
>   $ rm /var/tmp/kdecache-$USER/icon-cache.kcache
> 
> and restart KDE.

I guess not, I have no real clue about KDE, but when you look at kde4-base.eclass it also uses "gnome2_icon_cache_update" and inherits gnome2-utils.

[Comment 6 Julian Ospald 2012-08-23 19:57:07 UTC]

games.eclass has to go last, always. We want the default games functions exported and eclass behavior might change (like qt4-r2.eclass would suddenly export pkg_setup which will overwrite games_pkg_setup and break the toolchain exports) and could randomly compromise games ebuilds.

[Comment 7 Julian Ospald 2012-08-23 20:03:13 UTC]

also, you removed all runtime dependencies without knowing (by bumping to EAPI=4)

"When RDEPEND is unset, there will no longer be an automatic assignment of RDEPEND="${DEPEND}". "

from http://devmanual.gentoo.org/ebuild-writing/eapi/index.html

[Comment 8 Nikos Chantziaras 2012-08-24 05:28:55 UTC]

Indeed. I'll fix that. I'll do a 2.1.5 release first because 2.1.4 can't run TADS 3.1.2 games (there was a VM change) and will update the bump request with a corrected ebuild.

[Comment 9 Julian Ospald 2012-08-24 17:45:16 UTC]

I bumped 2.1.4 in the tree, open a new bug about 2.1.5 when it's ready

[Comment 10 Nikos Chantziaras 2012-09-04 11:02:01 UTC]

A few notes:

You're enforcing verbose builds. Is this a Gentoo standard I was not aware of? I see lots of packages that use terse build messages. This makes it easier to see warnings.

In make_desktop_entry, you missed the GenericName attribute. It should be:

make_desktop_entry qtads QTads qtads Game "GenericName=TADS Multimedia Interpreter\nMimeType=application/x-tads;application/x-t3vm-image;"

[Comment 11 Julian Ospald 2012-09-04 13:52:36 UTC]

(In reply to comment #10)
> A few notes:
> 
> You're enforcing verbose builds. Is this a Gentoo standard

yes

[Comment 12 Nikos Chantziaras 2012-09-07 14:21:25 UTC]

Btw, I fixed that issue in bug 433918 without having to use toolchain-funcs.