Re: [SCM] multicat packaging branch, master, updated. debian/1.0-1-3-g4c116f9

2010-08-25 Thread Reinhard Tartler
On Tue, Aug 24, 2010 at 23:05:46 (CEST), Jonas Smedegaard wrote:

 On Tue, Aug 24, 2010 at 10:24:25PM +0200, Rémi Duraffort wrote:
Le mardi 24 août 2010 à 09:33:56, Jonas Smedegaard a écrit :
 [sent again, to proper mailinglist this time]

 On Tue, Aug 24, 2010 at 07:08:32PM +,
 ivoire-gu...@users.alioth.debian.org wrote:

Some patches to add a missing license and fix some typos.

 diff --git a/debian/gbp.conf b/debian/gbp.conf
 index 8e96d07..1ee58fc 100644
 --- a/debian/gbp.conf
 +++ b/debian/gbp.conf
 @@ -1,3 +1,4 @@
 [git-buildpackage]
 pristine-tar = True
 compression = bzip2
 +ignore-new = True

 Above is not a patch.
You are right I might have split it in two commits.

 Also, I find it unwise to enable that option - if source is not clean
 after build + clean then something is wrong which should be fixed
 instead.
 I'm usually doing git-buildpackage, change something, commit it and
 then git-buildpackage again. The second one complain. It does not
 complain if I add this option (but I can do a dh_quilt_unpatch before
 git-buildpackage if this option is not recommended)

 git-buildpackage currently do not work 100% with quilt variant of source
 format 3.0.

That's why I was at first against adopting format 3.0 for this team.

 What I personally do after clean is this - manually:

   QUILT_PATCHES=debian/patches quilt pop -a
   [check that .pc is virtually empty]
   rm -rf .pc

 I dislike integrating above with packaging rules, as I consider the
 issue a bug/limitation in either or both of git-buildpackage and dpkg,
 so fixing it in packaging really means covering over a bug somewhere
 else.

indeed, and I at some point asked to integrate it into debian/rules to
streamline the workflow on Fromat 1.0 and Format 3.0 (quilt) packages.

 Others in this team disagree with me.

 Some git-ignore .pc subdir.

seems same to me, as dpkg ignores it as well per default (well, at least
with the option -i)

 Some implement quit-unrolling in clean rule or some other rule.

I think the clean rule is the right rule for that.

please note that if you do that, you also need to make sure additionally
that patches are being reapplied just before the build.

the obvious drawback is that patches are applied and unapplied unnecessarily.

-- 
Gruesse/greetings,
Reinhard Tartler, KeyID 945348A4

___
pkg-multimedia-maintainers mailing list
pkg-multimedia-maintainers@lists.alioth.debian.org
http://lists.alioth.debian.org/mailman/listinfo/pkg-multimedia-maintainers


Re: [SCM] multicat packaging branch, master, updated. debian/1.0-1-3-g4c116f9

2010-08-24 Thread Jonas Smedegaard

[sent again, to proper mailinglist this time]

On Tue, Aug 24, 2010 at 07:08:32PM +, ivoire-gu...@users.alioth.debian.org 
wrote:


   Some patches to add a missing license and fix some typos.

diff --git a/debian/gbp.conf b/debian/gbp.conf
index 8e96d07..1ee58fc 100644
--- a/debian/gbp.conf
+++ b/debian/gbp.conf
@@ -1,3 +1,4 @@
[git-buildpackage]
pristine-tar = True
compression = bzip2
+ignore-new = True


Above is not a patch.

Also, I find it unwise to enable that option - if source is not clean 
after build + clean then something is wrong which should be fixed 
instead.



  - Jonas

--
  * Jonas Smedegaard - idealist  Internet-arkitekt
  * Tlf.: +45 40843136  Website: http://dr.jones.dk/

  [x] quote me freely  [ ] ask before reusing  [ ] keep private


signature.asc
Description: Digital signature
___
pkg-multimedia-maintainers mailing list
pkg-multimedia-maintainers@lists.alioth.debian.org
http://lists.alioth.debian.org/mailman/listinfo/pkg-multimedia-maintainers


Re: [SCM] multicat packaging branch, master, updated. debian/1.0-1-3-g4c116f9

2010-08-24 Thread Rémi Duraffort
Le mardi 24 août 2010 à 09:33:56, Jonas Smedegaard a écrit :
 [sent again, to proper mailinglist this time]

 On Tue, Aug 24, 2010 at 07:08:32PM +, 
 ivoire-gu...@users.alioth.debian.org wrote:

Some patches to add a missing license and fix some typos.

 diff --git a/debian/gbp.conf b/debian/gbp.conf
 index 8e96d07..1ee58fc 100644
 --- a/debian/gbp.conf
 +++ b/debian/gbp.conf
 @@ -1,3 +1,4 @@
 [git-buildpackage]
 pristine-tar = True
 compression = bzip2
 +ignore-new = True

 Above is not a patch.
You are right I might have split it in two commits.

 Also, I find it unwise to enable that option - if source is not clean  
 after build + clean then something is wrong which should be fixed  
 instead.
I'm usually doing git-buildpackage, change something, commit it and then
git-buildpackage again. The second one complain. It does not complain if
I add this option (but I can do a dh_quilt_unpatch before
git-buildpackage if this option is not recommended)


-- 
Rémi Duraffort | ivoire
http://ivoire.dinauz.org/blog/

___
pkg-multimedia-maintainers mailing list
pkg-multimedia-maintainers@lists.alioth.debian.org
http://lists.alioth.debian.org/mailman/listinfo/pkg-multimedia-maintainers


Re: [SCM] multicat packaging branch, master, updated. debian/1.0-1-3-g4c116f9

2010-08-24 Thread Felipe Sateler
On 24/08/10 16:24, Rémi Duraffort wrote:
 Le mardi 24 août 2010 à 09:33:56, Jonas Smedegaard a écrit :
 [sent again, to proper mailinglist this time]

 On Tue, Aug 24, 2010 at 07:08:32PM +, 
 ivoire-gu...@users.alioth.debian.org wrote:

Some patches to add a missing license and fix some typos.

 diff --git a/debian/gbp.conf b/debian/gbp.conf
 index 8e96d07..1ee58fc 100644
 --- a/debian/gbp.conf
 +++ b/debian/gbp.conf
 @@ -1,3 +1,4 @@
 [git-buildpackage]
 pristine-tar = True
 compression = bzip2
 +ignore-new = True

 Above is not a patch.
 You are right I might have split it in two commits.
 
 Also, I find it unwise to enable that option - if source is not clean  
 after build + clean then something is wrong which should be fixed  
 instead.
 I'm usually doing git-buildpackage, change something, commit it and then
 git-buildpackage again. The second one complain. It does not complain if
 I add this option (but I can do a dh_quilt_unpatch before
 git-buildpackage if this option is not recommended)

It is not necessary to do git-buildpackage to test a change.
Also, to avoid this very problem, I set in my ~/.gbp.conf

export-dir=../build-area

So it will build in a clean directory, and not touch my working tree.

-- 
Saludos,
Felipe Sateler

___
pkg-multimedia-maintainers mailing list
pkg-multimedia-maintainers@lists.alioth.debian.org
http://lists.alioth.debian.org/mailman/listinfo/pkg-multimedia-maintainers


Re: [SCM] multicat packaging branch, master, updated. debian/1.0-1-3-g4c116f9

2010-08-24 Thread Jonas Smedegaard

On Tue, Aug 24, 2010 at 10:24:25PM +0200, Rémi Duraffort wrote:

Le mardi 24 août 2010 à 09:33:56, Jonas Smedegaard a écrit :

[sent again, to proper mailinglist this time]

On Tue, Aug 24, 2010 at 07:08:32PM +, 
ivoire-gu...@users.alioth.debian.org wrote:



   Some patches to add a missing license and fix some typos.

diff --git a/debian/gbp.conf b/debian/gbp.conf
index 8e96d07..1ee58fc 100644
--- a/debian/gbp.conf
+++ b/debian/gbp.conf
@@ -1,3 +1,4 @@
[git-buildpackage]
pristine-tar = True
compression = bzip2
+ignore-new = True


Above is not a patch.

You are right I might have split it in two commits.

Also, I find it unwise to enable that option - if source is not clean 
after build + clean then something is wrong which should be fixed 
instead.
I'm usually doing git-buildpackage, change something, commit it and 
then git-buildpackage again. The second one complain. It does not 
complain if I add this option (but I can do a dh_quilt_unpatch before 
git-buildpackage if this option is not recommended)


git-buildpackage currently do not work 100% with quilt variant of source 
format 3.0.


What I personally do after clean is this - manually:

  QUILT_PATCHES=debian/patches quilt pop -a
  [check that .pc is virtually empty]
  rm -rf .pc

I dislike integrating above with packaging rules, as I consider the 
issue a bug/limitation in either or both of git-buildpackage and dpkg, 
so fixing it in packaging really means covering over a bug somewhere 
else.


Others in this team disagree with me.

Some git-ignore .pc subdir.

Some implement quit-unrolling in clean rule or some other rule.

I believe that your above approach (I notice that you since reverted it) 
is bad even if you find my approach too complex/stupid/whatever: it 
ignores *any* changes in source, not only the patches.



Hope that helps.

 - Jonas

--
 * Jonas Smedegaard - idealist  Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private


signature.asc
Description: Digital signature
___
pkg-multimedia-maintainers mailing list
pkg-multimedia-maintainers@lists.alioth.debian.org
http://lists.alioth.debian.org/mailman/listinfo/pkg-multimedia-maintainers