Summary: | app-admin/gtkdiskfree: version 2.0.3 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Samuel Bauer <samuel.bauer> |
Component: | Current packages | Assignee: | Samuel Bauer <samuel.bauer> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | proxy-maint, xmw |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
gtkdiskfree-2.0.3.ebuild
gtkdiskfree-2.0.3.ebuild.patch metadata.xml.patch gtkdiskfree-2.0.3.ebuild gtkdiskfree-2.0.3.ebuild.diff |
Please make sure you comply with https://wiki.gentoo.org/wiki/Project:Proxy_Maintainers/User_Guide#Commit_Policy_for_Users. Your ebuild must pass repoman QA checks. As is, it will not. You need to update the copyright and header field at minimum. Regarding the drop of the nls USE flag, I haven't looked at the code from upstream, but is dropping the dependency correct? You now always depend on gettext, no? Created attachment 429804 [details, diff]
gtkdiskfree-2.0.3.ebuild.patch
sorry I'm not very regarding for headers, as it doesn't affect anything else than repoman.
About nls, last version passage didn't drop flag, but forced gettext without using flag as it is configure(d) by default, I dropped
Created attachment 429806 [details, diff]
metadata.xml.patch
Still about nls, maybe better to keep the flag, and the dependency management.
I introduced this problem, with 2.0.2 revision, just because I sent ebuild as it is in a personal overlay, and I sent maybe a bit fast.
(In reply to Samuel BAUER from comment #2) > About nls, last version passage didn't drop flag, but forced gettext without > using flag as it is configure(d) by default, I dropped (In reply to Samuel BAUER from comment #3) > Still about nls, maybe better to keep the flag, and the dependency > management. Keeping the flag itself doesn't make sense, as it is no longer toggle-able. The dependency on gettext though, should be permanent now, as it is effectively like nls is forced on for everyone, no? Created attachment 430000 [details]
gtkdiskfree-2.0.3.ebuild
I would tend to preserve the nls flag.
Thanks for the remark.
(In reply to Samuel BAUER from comment #5) > Created attachment 430000 [details] > gtkdiskfree-2.0.3.ebuild > > I would tend to preserve the nls flag. Why? > Thanks for the remark. What NP-Hardass is suggesting is DEPEND="${RDEPEND} virtual/pkgconfig sys-devel/gettext" If gettext is unconditionally required, it does not belong warrant being dependent upon an IUSE flag. Keeping the flag once unconditionally required is inconsistent with gentoo's ebuilding ruleset. > What NP-Hardass is suggesting is > > DEPEND="${RDEPEND} > virtual/pkgconfig > sys-devel/gettext" > > If gettext is unconditionally required, it does not belong warrant being > dependent upon an IUSE flag. Keeping the flag once unconditionally required is > inconsistent with gentoo's ebuilding ruleset. gettext isn't unconditionally required, one can --disable-nls Hope I didn't break this when modifying autotools, or gettextize, if it is let me know, it's a mistake. But when --disable-nls, it seems to me to have the correct behaviour I shouldn't submit the change from: $(use_enable nls); to: --enable-nls; in 2.0.2, as this lead me to drop the flag in 2.0.3 first submission, thinking it as an inconsistency in the ebuild. I really should not pick ebuilds from my personal overlay to send them here, as they may stick to my personal preferences. I have a "bugs" overlay I usually put anything related to the gentoo bugtracker, but for this one I was a bit too fast (In reply to Samuel BAUER from comment #7) > > What NP-Hardass is suggesting is > > > > DEPEND="${RDEPEND} > > virtual/pkgconfig > > sys-devel/gettext" > > > > If gettext is unconditionally required, it does not belong warrant being > > dependent upon an IUSE flag. Keeping the flag once unconditionally required is > > inconsistent with gentoo's ebuilding ruleset. > > gettext isn't unconditionally required, one can --disable-nls > Hope I didn't break this when modifying autotools, or gettextize, if it is > let me know, it's a mistake. But when --disable-nls, it seems to me to have > the correct behaviour > > I shouldn't submit the change from: $(use_enable nls); to: --enable-nls; in > 2.0.2, as this lead me to drop the flag in 2.0.3 first submission, thinking > it as an inconsistency in the ebuild. > > I really should not pick ebuilds from my personal overlay to send them here, > as they may stick to my personal preferences. I have a "bugs" overlay I > usually put anything related to the gentoo bugtracker, but for this one I > was a bit too fast Please paste your full build log your current ebuild. also, you have "$(use enable nls)" it needs to be $(use_enable nls) Just drop nls flag and include mandatory dependency. After looking at: https://developer.gnome.org/glib/stable/glib-I18N.html I thinks it's better to drop custom gettext macros and rely on glib ones. I'll fix this as soon as possible. Sorry having been so time wasting about this, and once again thanks for commenting. That's fine, we await a patch accordingly Created attachment 454502 [details, diff]
gtkdiskfree-2.0.3.ebuild.diff
Let's get this going. Builds fine with GCC-5.3.0.
commit e0a99c87671c0049a6867164fdfaaab792c21928 (HEAD) Author: Andreas Sturmlechner <andreas.sturmlechner@gmail.com> AuthorDate: Mon Dec 5 22:55:58 2016 +0100 Commit: Patrice Clement <monsieurp@gentoo.org> CommitDate: Wed Dec 7 22:48:56 2016 +0100 app-admin/gtkdiskfree: version bump to 2.0.3 and fix build w/ GCC-5. Gentoo-Bug: https://bugs.gentoo.org/578768 Gentoo-Bug: https://bugs.gentoo.org/586956 Package-Manager: portage-2.3.0 Closes: https://github.com/gentoo/gentoo/pull/3025 app-admin/gtkdiskfree/Manifest | 1 + app-admin/gtkdiskfree/gtkdiskfree-2.0.3.ebuild | 29 ++++++++++++++++++++++++++ 2 files changed, 30 insertions(+) create mode 100644 app-admin/gtkdiskfree/gtkdiskfree-2.0.3.ebuild PR merged, thanks! Thanks Patrice Clement. |
Created attachment 429440 [details] gtkdiskfree-2.0.3.ebuild do not include --enable-nls from econf, as it is default, and removed unused nls flag. added a old-color-selector flag: allows to use the old good but "deprecated" GtkColorSelectionDialog, even compiling for gtk3 (default for gtk2). There may be a shorter name for this flag. The last warning about deprecation in this application. I would not consider stabilizing 2.0.2, as in lower or equal versions, capacity images, and column sort leaked (under 2.0.2 it even froze the app entering infinite loop). Moreover 2.0.3 offers new features and "modernizes" interface. I