Re: [Git-dpm-user] Repackaging roundup

2013-06-18 Thread Bernhard R. Link
* Kai Storbeck  [130617 23:24]:
> As an exercise I ran git-dpm linearize to remove the first patch that
> removes COPYING.txt from the root.
> 
> So far so good, "Re'base' -i seems to have been successfull". After
> running git-dpm update-patches, the file COPYING.txt is still missing.
> This file should reappear in the "applied-patches" state, iiuc?

That seems to be a little inconsistency caused by the deleted file
detection code. In a case like that you need to call update-patches
(or to be more precise merge-patched, though that is usually called via
update-patches) with the --ignore-deletions switch.
(if that merge was your last commit, you can regenerate it with
git-dpm checkout-patched ; git checkout master ; git reset --hard HEAD^1
git-dpm update-patches --ignore deletions).

The issue is that without that option git-dpm will keep files deleted
that were deleted in your Debian branch before. (So that one can just
commit the changes a "debian/rules clean" would do to the Debian
branch). The problem is that git-dpm compares the old upstream branch
with the old Debian branch to look for files deleted in the Debian
branch, which fails with files being deleted in the patched branch.
(and the code in question only knows the old upstream branch not the old
patched branch, I'll have to take a look if the information about the
old patched branch can be made available there).

> (Minor detail: Successful is spelled with a single l, attached a patch)

Thanks. English is not my first language and especially that dropping
and readding letters is something I too often get wrong.

Bernhard R. Link

___
Git-dpm-user mailing list
Git-dpm-user@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/git-dpm-user


Re: [Git-dpm-user] Repackaging roundup

2013-06-17 Thread Kai Storbeck
Hi Bernhard,

Thanks for your reply.

On 6/12/13 7:22 PM, Bernhard R. Link wrote:
> * Kai Storbeck  [130612 18:34]:
>> Is the master branch in a pre-applied stage? Or in an applied stage?
> 
> The master branch is supposed to be in a applied branch. The idea is
> that git checkout and a dpkg-source -x should produce comparable
> results and something one can just call dpkg-buildpackage in.
> 
>> Am I hitting a bug, are the warnings harmless?
> 
> Those warnings are harmless. I think they are a minor bug in dpkg
> triggered by your first patch only removing files. (dpkg tries to
> detect whether all patches are properly applied by trying to
> apply them and the first patch triggers an warning instead of an
> error so that is output).

Okay.

> Note that removing those files via a patch is quite a unusual and
> wasteful way (they are fully included in the diff a second time).
> It might be better to remove them at the proper place.

I was pointed out by others that this is a weird route to take. I'd like
to clean that up for my next upload for roundup.

> What is a bug is that git-dpm fails to parse Author: 
> in those patches (without any name given). It might make sense
> to run git-dpm linearize and ammending those commits to get the
> right author and not the Author: artefact there. (fixed in git-dpm's
> git repository in case you want to reimport it instead).

Neat.

I have reimported it with the git version.

As an exercise I ran git-dpm linearize to remove the first patch that
removes COPYING.txt from the root.

So far so good, "Re'base' -i seems to have been successfull". After
running git-dpm update-patches, the file COPYING.txt is still missing.
This file should reappear in the "applied-patches" state, iiuc?

(Minor detail: Successful is spelled with a single l, attached a patch)

Could you nudge me a bit more ? Perhaps I'm missing some essential
git{,-dpm} knowledge?

Regards,
Kai

From 84d7de6fa7a40ee2e0dcc260605237d5cecee35a Mon Sep 17 00:00:00 2001
From: Kai Storbeck 
Date: Mon, 17 Jun 2013 23:19:05 +0200
Subject: fix a minor spelling mistake

---
 git-dpm.sh|4 ++--
 tests/test.sh |2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/git-dpm.sh b/git-dpm.sh
index acea97e..0b7d718 100755
--- a/git-dpm.sh
+++ b/git-dpm.sh
@@ -1879,7 +1879,7 @@ EOF
echo "Nothing changed. (Did you change your mind?)"
 
else
-   echo "Re'base' -i seems to have been successfull. Don't forget 
to call update-patches!"
+   echo "Re'base' -i seems to have been successful. Don't forget 
to call update-patches!"
# TODO: perhaps call it here?
# (but don't forget to update PATCHEDREV first...)
fi
@@ -1972,7 +1972,7 @@ function rebase_patches() {
echo "Don't forget to call update-patches, though."
fi
else
-   echo "Rebase seems to have been successfull. Don't forget to 
call update-patches!"
+   echo "Rebase seems to have been successful. Don't forget to 
call update-patches!"
PATCHEDREV="$(gitcmd rev-parse --verify -q HEAD)" || return 1
fi
return 0
diff --git a/tests/test.sh b/tests/test.sh
index 62f141f..33bcc6a 100755
--- a/tests/test.sh
+++ b/tests/test.sh
@@ -971,7 +971,7 @@ fi #specialbranchnames
 
 if $sometestsfound ; then
echo "# END 
###"
-   echo "$testtodo tests successfull..."
+   echo "$testtodo tests successful..."
 else
echo "Unknown test $testtodo"
exit 1
-- 
1.7.10.4



signature.asc
Description: OpenPGP digital signature
___
Git-dpm-user mailing list
Git-dpm-user@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/git-dpm-user

Re: [Git-dpm-user] Repackaging roundup

2013-06-12 Thread Bernhard R. Link
* Kai Storbeck  [130612 18:34]:
> Is the master branch in a pre-applied stage? Or in an applied stage?

The master branch is supposed to be in a applied branch. The idea is
that git checkout and a dpkg-source -x should produce comparable
results and something one can just call dpkg-buildpackage in.

> Am I hitting a bug, are the warnings harmless?

Those warnings are harmless. I think they are a minor bug in dpkg
triggered by your first patch only removing files. (dpkg tries to
detect whether all patches are properly applied by trying to
apply them and the first patch triggers an warning instead of an
error so that is output).
Note that removing those files via a patch is quite a unusual and
wasteful way (they are fully included in the diff a second time).
It might be better to remove them at the proper place.

What is a bug is that git-dpm fails to parse Author: 
in those patches (without any name given). It might make sense
to run git-dpm linearize and ammending those commits to get the
right author and not the Author: artefact there. (fixed in git-dpm's
git repository in case you want to reimport it instead).

Bernhard R. Link

___
Git-dpm-user mailing list
Git-dpm-user@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/git-dpm-user