On May 1, 2015, at 9:08 AM, Stuart Henderson <[email protected]> wrote:

> On 2015/05/01 17:07, Stuart Henderson wrote:
>> On 2015/05/01 08:36, Eric Lalonde wrote:
>>>> New build dependencies are necessary: e2fsprogs (uuid-related)
>> 
>> What's the story with this? It doesn't use e2fsprogs libuuid at all,
>> there are #include <uuid/uuid.h> in a couple of files but they don't
>> use anything from it, there is no problem if these are just removed.
>> 
> 

Thank you for your review. audio/jack started using libuuid/uuid.h after 
0.121.3 but stopped using it before 0.124.1. The now-unnecessary #include lines 
were not removed upstream until after 0.124.1. I have added a patch to remove 
those lines, which itself can be deleted once upstream releases the next 
version.

> Oh, there was one other thing -
> 
> echo "#define JACKD_MD5_SUM \"`md5sum .libs/jackd | awk '{print $1}'`\"" > 
> jack_md5.h
> /bin/sh: md5sum: not found

This ended up being harmless, and an issue which pre-dates this diff: 
jack_md5.h is included when building the jackstart utility, but jackstart is 
only built when  --enable-capabilities is specified (which we don’t, it 
wouldn’t work). The autoconf scripts are smart enough not to test for md5sum 
unless capabilities are enabled, but not smart enough to only generate 
jack_md5.h if jackstart is being built. For the sake of avoiding the display of 
spurious errors, I have included a patch to generate jack_md5.h properly on an 
OpenBSD system, even though we don’t build jackstart. I’ll look into 
communicating this issue upstream.

New diff attached. Thanks for the review.

- Eric

Attachment: jack-0.124.1.diff
Description: Binary data


Reply via email to