svn commit: r320973 - head/audio/grip

[Baptiste Daroussin]

On Sat, Jun 15, 2013 at 02:18:48PM +0400, Boris Samorodov wrote:
> 15.06.2013 13:20, Alexey Dokuchaev пишет:
> > On Sat, Jun 15, 2013 at 01:06:06PM +0400, Boris Samorodov wrote:
> >> The shlib ABI version was removed from LIB_DEPENDS and it was not
> >> reflected at the log. So the question arises if it was intended.
> >> Imho (yep, it's really imho) this change is serious enough to go
> >> to the log.
> > 
> > While I support you here on necessity of documenting every change in the
> > commit log, dropping ABI versions from LIB_DEPENDS had been common practice
> > for a while now, and can be forgiven.
> 
> -----
>  LIB_DEPENDS=	id3-3.8.3:${PORTSDIR}/audio/id3lib \
> -		curl.6:${PORTSDIR}/ftp/curl
> +		curl:${PORTSDIR}/ftp/curl
> -----
> 
> Well, I'm not sure if the change was intended and it was not
> documented. Hence my question. And why curl was changed while
> id3 -- not? (BTW it's just a rhetoric question)

I removed the shlib for curl because is it safe and I didn't for id3 because it
will hit the buggy LIB_DEPENDS I'm speaking in my previous CFT mail.

id3-3.8 will match libid3-3.so.8 and libid3-3.8.so so I kept the shlib number.

> 
> I don't blame Baptiste. I was just curious if that change was intended.
> And as a side note, if that change was logged, the question wouldn't
> occur. As simple as that.

I don't feel blamed and I do like reviews and questions, that allows to double
check something done is right or wrong :).

> 
> >  We have much worse examples of
> > commits documented with one-liner log (which is bogus most of the times);
> > now, *that* is a problem. :(
> 
> +1 :-(

Agree I will try to improve my commit logs.

> 
> >> Second, the package definitely had changed. The dependency had
> >> changed from libcurl.so.6 to libcurl.so. And this is recorded
> >> at the package.
> > 
> > I didn't know we're recording fully expanded shlibs in +CONTENTS.  Library
> > versions are recorded inside the binaries,
> 
> Yes, that means that a package (one or more binaries) has changed.

No this information is only concerning the package neither the binary, it is a
port framework only information.
> 
> > but in this case, dropping them
> > from LIB_DEPENDS should be no-op.
> 
> Is it because the dependency is more wide? If yes, than thanks, I need
> just this information: "It is not a significant change for the package."

This change gives exactly no change at all to the generated package neither to
the binary.

regards,
Bapt
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/svn-ports-all/attachments/20130615/e672cc37/attachment.sig>