On Sat, Jan 04, 2014 at 07:11:58PM +0100, Stefan Fritsch wrote:
> You are right, the real fix is not included. We "fixed" the issue by
> removing the patches that touch *.m4 files. Therefore, no *.m4 file
> will be created in the .pc directory.
>
> This reminds me that I need to propose the upstream patch for backport
> to 2.4.8. Maybe we should put the fix into the debian package for
> 2.4.7-2, too, but there are usually few reasons to patch the m4 files,
> now that we have moved mpm-itk to its own source package.
I'm wondering if there's more of a potential issue in Ubuntu than in
Debian here. Our next release will be supported for five years, so I
expect that we'll backport a number of as-yet-unknown bug fixes. As
we're team based, it could be anybody doing this, and they may be
unaware of this issue.
So I think I'd prefer to carry the backported fix for this, just so
there isn't a mine for a future developer to step on.
We prefer to keep our delta against Debian small, so that's a vote from
me to have the fix in 2.4.7-2, please. That's assuming that we just need
the ignore-quilt-dir patch, or are there any complications with
build system regeneration or anything like that?
I'll also look to see if I can write a dep8 test that just checks for
duplicate module loads.
Hi Stefan,
Thank you for explaining this to me.
On Sat, Jan 04, 2014 at 07:11:58PM +0100, Stefan Fritsch wrote:
> You are right, the real fix is not included. We "fixed" the issue by
> removing the patches that touch *.m4 files. Therefore, no *.m4 file
> will be created in the .pc directory.
>
> This reminds me that I need to propose the upstream patch for backport
> to 2.4.8. Maybe we should put the fix into the debian package for
> 2.4.7-2, too, but there are usually few reasons to patch the m4 files,
> now that we have moved mpm-itk to its own source package.
I'm wondering if there's more of a potential issue in Ubuntu than in
Debian here. Our next release will be supported for five years, so I
expect that we'll backport a number of as-yet-unknown bug fixes. As
we're team based, it could be anybody doing this, and they may be
unaware of this issue.
So I think I'd prefer to carry the backported fix for this, just so
there isn't a mine for a future developer to step on.
We prefer to keep our delta against Debian small, so that's a vote from
me to have the fix in 2.4.7-2, please. That's assuming that we just need
the ignore-quilt-dir patch, or are there any complications with
build system regeneration or anything like that?
I'll also look to see if I can write a dep8 test that just checks for
duplicate module loads.
Thanks,
Robie