Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 16 Jun 2004 10:00:43 GMT
From:      Alexander Nedotsukov <bland@FreeBSD.org>
To:        gnome@FreeBSD.org
Subject:   Re: ports/67970: ports textproc/libxml, textproc/libxslt: bogus dependencies on devel/pkgconfig
Message-ID:  <200406161000.i5GA0hcA013571@freefall.freebsd.org>

next in thread | raw e-mail | index | archive | help
The following reply was made to PR ports/67970; it has been noted by GNATS.

From: Alexander Nedotsukov <bland@FreeBSD.org>
To: freebsd-gnats-submit@FreeBSD.org, eikemeier@fillmore-labs.com
Cc:  
Subject: Re: ports/67970: ports textproc/libxml, textproc/libxslt: bogus dependencies
 on devel/pkgconfig
Date: Wed, 16 Jun 2004 18:52:55 +0900

 Oliver, ports which installs *.la files installs them under existing 
 hierarchy (and btw they better not to install them at all). Let's try on 
 clean machine pkg_add some; pkg_delete pkgconfig; pkg_delete some; Why 
 empty libdata/pkgconfig leftover? Because we don't have something like 
 installation reference counting in Solaris and run-time dependency fixes 
 the situation here. This actually answer for your point two from 
 original posting also.
 
 Now argument three from the same place. I beleive in most of the cases 
 users do not requeired to rebuild ports which depend on pkgconfig 
 because of version bump in last one. Why do you think they have to?
 
 Add here my feeling that argument one is pretty much artificial. Users 
 will fetch, build and install pkgconfig anyway. Well not at the libxml 
 build time (with your proposed solution) but I bet just after it. The 
 reason is simple. For the user (not developer) libxml itself have zero 
 value. And when s/he start to build port which depends on libxml 
 pkgconfig requirement will be unavoidable.
 
 For my understanding the right way how situation may be fixed is to 
 patch bsd.gnome.mk and add fake-package which un/install 
 libdata/pkgconfig.  This will complicate things a bit but do not lower 
 dependency lists though. We only get small advantage in size <40K overall.
 
 Btw libxml just one case of many we already have. Do you realize this?
 



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200406161000.i5GA0hcA013571>