I have very limited knowledge in this area. Given Alan's analysis, I
agree with the first alternative as well.
Thanks,
George
On 12/11/2011 10:21 AM, Alan Coopersmith wrote:
On 12/11/11 10:14, Rich Burridge wrote:
Hi,
I've seen two different people on two different lists trip over this
one recently, so let's fix it.
Could I please get a code review for my fix for:
7088570 gcc packages should have dependencies on system/header
http://monaco.us.oracle.com/detail.jsf?cr=7088570
Webrev is at:
http://jurassic.us.oracle.com/~richb/7088570-v1/
LGTM, though I'm sure you'll get more opinions on Monday.
We could fix this with:
depend fmri=pkg:/system/header type=require
(which is the approach I took), or with something like:
depend fmri=__TBD pkg.debug.depend.file=usr/include/stdio.h type=require
I prefer the first, since it's unversioned, while the second is likely to
generate a dependency on the version installed on the build machine, which
is unnecessarily restrictive - any version of system/header should be
sufficient to install the base C language header files, and we shouldn't
force an ON upgrade on users who want gcc fixes.
_______________________________________________
userland-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/userland-discuss
