1:33 p.m.
I've updated the iconcache proposal:
http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
per the suggestions made at the recent fedora-packaging meeting.
In short, simplify to use xdg-utils, and add (when needed):
Requires(post): xdg-utils
Requires(postun): xdg-utils
We could likely optimize %postun to include
if [ $1 -eq 0 ]; then
...
fi
but I'm not sure if it's worth it.
-- Rex
11:37 a.m.
On Wed, 2006-12-13 at 13:33 -0600, Rex Dieter wrote:
I've updated the iconcache proposal:
http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
per the suggestions made at the recent fedora-packaging meeting.
In short, simplify to use xdg-utils, and add (when needed):
Requires(post): xdg-utils
Requires(postun): xdg-utils
We could likely optimize %postun to include
if [ $1 -eq 0 ]; then
...
fi
but I'm not sure if it's worth it.
Hey Rex,
Looks like your edit got lost somewhere. The last edit I see to the
page was 2006-12-04.
-Toshio
11:51 a.m.
Toshio Kuratomi wrote:
On Wed, 2006-12-13 at 13:33 -0600, Rex Dieter wrote:
> I've updated the iconcache proposal:
> http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
> per the suggestions made at the recent fedora-packaging meeting.
>
> In short, simplify to use xdg-utils, and add (when needed):
> Requires(post): xdg-utils
> Requires(postun): xdg-utils
>
> We could likely optimize %postun to include
> if [ $1 -eq 0 ]; then
> ...
> fi
> but I'm not sure if it's worth it.
Hey Rex,
Looks like your edit got lost somewhere. The last edit I see to the
page was 2006-12-04.
Arg! You're right. ????
ok, one more try.
-- Rex
2:53 p.m.
Rex Dieter wrote:
Toshio Kuratomi wrote:
> Hey Rex,
> Looks like your edit got lost somewhere. The last edit I see to the
> page was 2006-12-04.
Arg! You're right. ????
OK,
http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
*really* is v0.2 now.
To re-iterate, simplify to use xdg-utils, and add (when needed):
Requires(post): xdg-utils
Requires(postun): xdg-utils
4:21 p.m.
On Wednesday 13 December 2006 20:33, Rex Dieter wrote:
In short, simplify to use xdg-utils, and add (when needed):
Requires(post): xdg-utils
Requires(postun): xdg-utils
How about removing this completely from .spec files and add it to a
yum-plugin, that runs whenever package adds a file to {_datadir}/icons/ and
then only once per yum invocation?
Regards,
Till
5:23 p.m.
Till Maas wrote:
On Wednesday 13 December 2006 20:33, Rex Dieter wrote:
> In short, simplify to use xdg-utils, and add (when needed):
> Requires(post): xdg-utils
> Requires(postun): xdg-utils
How about removing this completely from .spec files and add it to a
yum-plugin, that runs whenever package adds a file to {_datadir}/icons/ and
then only once per yum invocation?
sure, got code for that? (:
Besides, I don't think wouldn't work for manual rpm installs.
-- Rex
5:42 p.m.
On Wed, Dec 13, 2006 at 01:33:22PM -0600, Rex Dieter wrote:
I've updated the iconcache proposal:
http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
per the suggestions made at the recent fedora-packaging meeting.
In short, simplify to use xdg-utils, and add (when needed):
Requires(post): xdg-utils
Requires(postun): xdg-utils
Hi,
I have two questions (which will have been answered, but I haven't
caught up with all traffic on this topic, so please answer again :):
a) "If none of the package's existing dependencies themselves already
depend on xdg-utils3, include ..."
I wouldn't rely on dependencies providing dependencies. Sure, we do
remove some redundancy technically, but cut&paste methods are used
more often than reading the guidelines, recursive dependencies may
change and so on. Let's keep it simple and always require it.
b) "someday when xdg-utils becomes universally available (hopefully,
this will include F*7),"
While the xdg-utils sound like a trivial tool the sentence seems to
imply that there are larger obstacles to getting this done. Why? If
this improves/simplifies package quality then who would block this?
--
Axel.Thimm at ATrpms.net
5:45 p.m.
On Fri, Dec 15, 2006 at 12:42:20AM +0100, Axel Thimm wrote:
On Wed, Dec 13, 2006 at 01:33:22PM -0600, Rex Dieter wrote:
> I've updated the iconcache proposal:
> http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
> per the suggestions made at the recent fedora-packaging meeting.
>
> In short, simplify to use xdg-utils, and add (when needed):
> Requires(post): xdg-utils
> Requires(postun): xdg-utils
Hi,
I have two questions (which will have been answered, but I haven't
caught up with all traffic on this topic, so please answer again :):
a) "If none of the package's existing dependencies themselves already
depend on xdg-utils3, include ..."
I wouldn't rely on dependencies providing dependencies. Sure, we do
remove some redundancy technically, but cut&paste methods are used
more often than reading the guidelines, recursive dependencies may
change and so on. Let's keep it simple and always require it.
b) "someday when xdg-utils becomes universally available (hopefully,
this will include F*7),"
While the xdg-utils sound like a trivial tool the sentence seems to
imply that there are larger obstacles to getting this done. Why? If
this improves/simplifies package quality then who would block this?
And the third of the two questions is: ;)
c) why is xdg-icon-resource's exit code thrown away? We do know it
will exist (explicit direct or indirect Requires:) contrary to
gtk-update-icon-cache, so if it fails it probably indicates some
system error that should propagate up to the user and get hos
attention (e.g. write failure to the cache/fs full etc.)
--
Axel.Thimm at ATrpms.net
2:27 a.m.
On Thu, Dec 14, 2006 at 06:27:07PM -0600, Rex Dieter wrote:
Axel Thimm wrote:
> c) why is xdg-icon-resource's exit code thrown away? We do know it
> will exist (explicit direct or indirect Requires:) contrary to
> gtk-update-icon-cache, so if it fails it probably indicates some
> system error that should propagate up to the user and get hos
> attention (e.g. write failure to the cache/fs full etc.)
failed scriptlets = aborted rpm transaction, which is bad bad bad.
Yes, I'm aware of that, aborted transaction = user is notified of
somethng broken. So the question is if xdg-icon-resource is always
present and is known (is it?) to not fail unless the file system was
remounted read-only or the binary is of the wrong elf format or other
catastrophies, then better fail early than late.
Of course the actual work not having been done by this script is
negligible, but I just don't want to silence away any failures in all
scriplets for the sake of (presumably) allowing rpm to install
further on.
--
Axel.Thimm at ATrpms.net
6:30 p.m.
Axel Thimm wrote:
On Fri, Dec 15, 2006 at 12:42:20AM +0100, Axel Thimm wrote:
> On Wed, Dec 13, 2006 at 01:33:22PM -0600, Rex Dieter wrote:
>> I've updated the iconcache proposal:
>> http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
>> per the suggestions made at the recent fedora-packaging meeting.
>>
>> In short, simplify to use xdg-utils, and add (when needed):
>> Requires(post): xdg-utils
>> Requires(postun): xdg-utils
> Hi,
>
> I have two questions (which will have been answered, but I haven't
> caught up with all traffic on this topic, so please answer again :):
>
> a) "If none of the package's existing dependencies themselves already
> depend on xdg-utils3, include ..."
>
> I wouldn't rely on dependencies providing dependencies. Sure, we do
...
>
> b) "someday when xdg-utils becomes universally available (hopefully,
> this will include F*7),"
>
> While the xdg-utils sound like a trivial tool the sentence seems to
> imply that there are larger obstacles to getting this done. Why? If
> this improves/simplifies package quality then who would block this?
blockers? None, that I'm aware of.
Well, a) is a just pre-cursor to b). I'd like to someday not need the
Requires: xdg-utils
at all.
I'm just as ok with Requiring it's unconditional use too.
-- Rex
2:29 a.m.
On Thu, Dec 14, 2006 at 06:30:24PM -0600, Rex Dieter wrote:
Axel Thimm wrote:
> On Fri, Dec 15, 2006 at 12:42:20AM +0100, Axel Thimm wrote:
>> On Wed, Dec 13, 2006 at 01:33:22PM -0600, Rex Dieter wrote:
>>> I've updated the iconcache proposal:
>>> http://fedoraproject.org/wiki/PackagingDrafts/ScriptletSnippets/iconcache
>>> per the suggestions made at the recent fedora-packaging meeting.
>>>
>>> In short, simplify to use xdg-utils, and add (when needed):
>>> Requires(post): xdg-utils
>>> Requires(postun): xdg-utils
>> Hi,
>>
>> I have two questions (which will have been answered, but I haven't
>> caught up with all traffic on this topic, so please answer again :):
>>
>> a) "If none of the package's existing dependencies themselves already
>> depend on xdg-utils3, include ..."
>>
>> I wouldn't rely on dependencies providing dependencies. Sure, we do
...
>>
>> b) "someday when xdg-utils becomes universally available (hopefully,
>> this will include F*7),"
>>
>> While the xdg-utils sound like a trivial tool the sentence seems to
>> imply that there are larger obstacles to getting this done. Why? If
>> this improves/simplifies package quality then who would block this?
blockers? None, that I'm aware of.
Then let's vote on it.
Well, a) is a just pre-cursor to b). I'd like to someday not
need the
Requires: xdg-utils
at all.
I'm just as ok with Requiring it's unconditional use too.
I'm fine with the proposal as is (wouldn't mind adding strict
requirements on xdg-utils3 and removing "|| :", but that's just nice
to have ;), so +1 from here
--
Axel.Thimm at ATrpms.net
6334
days inactive
6336
days old
packaging@lists.fedoraproject.org
11 comments
4 participants
participants (4)
-
Axel Thimm -
Rex Dieter -
Till Maas -
Toshio Kuratomi