From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:49048) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dCDMw-0001Sg-8C for guix-patches@gnu.org; Sat, 20 May 2017 19:01:07 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dCDMt-000596-61 for guix-patches@gnu.org; Sat, 20 May 2017 19:01:06 -0400 Received: from debbugs.gnu.org ([208.118.235.43]:54780) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1dCDMs-00058v-Jc for guix-patches@gnu.org; Sat, 20 May 2017 19:01:03 -0400 Received: from Debian-debbugs by debbugs.gnu.org with local (Exim 4.84_2) (envelope-from ) id 1dCDMs-0005gK-8m for guix-patches@gnu.org; Sat, 20 May 2017 19:01:02 -0400 Subject: bug#26988: Add desktop-file-utils to all gtk applications Resent-Message-ID: From: Marius Bakke In-Reply-To: <20170520150411.21c3e0aa@scratchpost.org> References: <20170518230111.53a91ffe@scratchpost.org> <87inkwa1n1.fsf@fastmail.com> <20170520150411.21c3e0aa@scratchpost.org> Date: Sun, 21 May 2017 01:00:22 +0200 Message-ID: <87h90f5e2h.fsf@fastmail.com> MIME-Version: 1.0 Content-Type: multipart/signed; boundary="=-=-="; micalg=pgp-sha512; protocol="application/pgp-signature" List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-patches-bounces+kyle=kyleam.com@gnu.org Sender: "Guix-patches" To: Danny Milosavljevic , =?UTF-8?Q?=E5=AE=8B=E6=96=87=E6=AD=A6?= Cc: 26988@debbugs.gnu.org --=-=-= Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Danny Milosavljevic writes: > Hi Marius, > Hi =E5=AE=8B=E6=96=87=E6=AD=A6, > > On Sat, 20 May 2017 01:05:22 +0200 > Marius Bakke wrote: > >> Danny Milosavljevic writes: >>=20 >> > these are two patches I propose we add in order to add >> > desktop-file-utils to all gtk applications. This means that the >> > profile hook would pick up the desktop files (as soon as possible).=20= =20 >>=20 >> So if I read (guix profiles) correctly, the hook only runs when >> 'desktop-file-utils' is referenced. Would checking for a gtk reference >> as well not achieve the same effect? Or is that too expensive? > > That would probably work as well. > > But the idea is KDE would eventually use the XDG desktop cache, too (mayb= e it already does - no idea). Any other desktop environments that don't re= invent the wheel in that regard would use desktop-file-utils as well - and = that's the package that actually contains the "update-desktop-database" exe= cutable for updating the XDG desktop cache. > > Right now the "xdg-desktop-database" profile hook doesn't need its own ma= gical packages but it just uses the existing desktop-file-utils reference f= or actually updating the desktop database. IIUC, the idea is to run the hook every time a package with a ".desktop" file has changed. Even if we propagate "desktop-file-utils" from say, GTK+ and Qt, it won't catch some cases where we add the ".desktop" file manually such as "rxvt-unicode". I wonder if there is a more generic thing we could test for.. Maybe libx11? Obviously lots of false positives, but the hook is not *that* expensive. Thoughts? >> > One patch moves desktop-file-utils from (gnu packages gnome) to (gnu >> > packages freedesktop). I suggest we apply this patch to all branches >> > - since they don't change anything in the build.=20=20 >>=20 >> Adding it to 'master' would be sufficient, no? It will be merged to >> 'staging' and 'core-updates' in due time. > >> I don't see a good reason to move it though, but no strong opinion.=20 > > desktop-file-utils is a freedesktop project. > > If we don't move it and if KDE used desktop-file-utils (as it should), we= 'd have KDE depend on Gnome. Ah, ok. Makes sense. > If we did move it in all branches, I mean also in wip-installer etc, I th= ink it would prevent any merging problems - since git uses file hashes as c= ommits - and people would not have to take care whether this branch has alr= eady moved it or not. IME, git usually recognizes cherry-picks as the same change, but not always. I think the problem is that "cherry-pick" is smart, it can adapt to some fairly complicated context changes unlike e.g. "patch". So, even if the cherry-pick succeeds, the file and context can be very different. For example, I recently tested merging a branch with 600+ cherry-picks back to 'master'. That went mostly-well, but there were some very difficult conflicts in all files that had changed in the other branch (and the duplicate branch history did not help). In short, I think we should prefer merging to cherry-picks whenever possible. Merging is equally smart, so if cherry-pick works "out of the box" merge will too. And they definitely do not prevent merging problems, quite the opposite! --=-=-= Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- iQEzBAEBCgAdFiEEu7At3yzq9qgNHeZDoqBt8qM6VPoFAlkgyoYACgkQoqBt8qM6 VPoIGAf9FJF29v2UKiAvYX29k02Lq1PjToF1itdyzVXFW62HcbSIw6bdD5bG99tm vAdzenpiULQwhYxo4ISDme1qUnWl7HmUZSKiPvMBWeDqYamIMrhDZEmX2yNm+1dW JR6yHF9MFHIUL9etyvSJCGZowLSJdGwmulEV47Ihdpk14jF+GAeADW7JaP67Bye2 BtWeNu7zRPuNpB7EpGjbJXSWM0cH2KD55oc3dp1VuwjryJap8a/GtBxnNaUACj/M beZ8ByTXKA9xjdAi+feyNAp8dZRK2l5x//7S9UDQ+tkgw2fLtit6C/eEqAzUZn0z ob2v1oIhNVZlILa67f225XElLJyjMw== =GnQB -----END PGP SIGNATURE----- --=-=-=--