On Wed, Oct 27, 2010 at 1:57 PM, Asheesh Laroia <ashe...@asheesh.org> wrote:
> On Tue, 26 Oct 2010, Evan Broder wrote:
>
>> Package: alpine
>> Version: 2.01-1
>> Severity: normal
>> Tags: patch
>>
>> alpine uses symbols from libkrb5 and libcrypto, but does not link
>> those libraries directly.
>>
>> Currently this does not cause a build failure because libkrb5 is
>> linked as a result of linking libgssapi, and libcrypto is linked by
>> linking libssl.
>>
>> However, it does cause build failures on Ubuntu Natty due to changes
>> in the linker configuration, and it will cause similar failures in
>> Wheezy for the same reason (see
>> http://wiki.debian.org/ToolChain/DSOLinking)
>>
>> I've attached a patch that fixes the issue in a fairly minimal way as
>> a straw man.
>>
>> Please let me know if you're not planning to upload fixes during the
>> Squeeze freeze, so that I can go ahead and fix this in Ubuntu.
>
> I don't know where you got "alpine 2.01" in your version string. But this is
> almost definitely a good idea to change.

Ah, yeah sorry - that was simply a typo on my part (my Debian machine
with a working sendbug is different from my Ubuntu machine where I
generate patches)

> Your patch is kind of excessive -- I don't tthink you really mean to do
> this:
>
> + case "$host" in
> +-  *-linux-gnu*|*-k*bsd*-gnu*|*-*-gnu*)
> ++  *-linux-gnu*|*-k*bsd*-gnu*)
>
> But in general I'm kind of struggling to understand this patch format where
> there's two nested patches.
>
> The patches aren't in a good format to go upstream. Can you improve them so
> they can be? You're changing configure.ac but then also patching configure;
> best to just re-run autogen.sh if possible.
>
> I'll take a closer look at this on Monday, but that's my thoughts for now.

Yes, sorry. I was targetting a fix in the packaging, so that debdiff
adds another quilt patch. I've attached a simpler patch that only
makes the configure.ac change. The unnecessary changes are probably
autoconf versioning differences.

> -- Asheesh.

- Evan

Attachment: alpine_linking.diff
Description: Binary data

Reply via email to