Tim Foster wrote:
> On Sun, 2010-08-08 at 20:31 -0700, Alan Coopersmith wrote:
>> Tim Foster wrote:
>>> * legacy action checking should probably report problems like the one
>>> below as warnings, not errors: there's a lot of these - we've never had
>>> individual legacy packages for a lot of packages. The combination of
>>> user/root packages also contribute to the noise here.
>>>
>>> ERROR pkglint.action003.2         unable to find ancestor
>>> SUNWespeak-devel marked in the 'pkg' attr of legacy action in
>>> pkg://opensolaris.org/library/speech/[email protected],5.11-0.145:20100730T020213Z
>> What does that error mean?   I've not heard of "ancestors" for legacy actions
>> before.
> 
> This was the assertion in the pkglint-plan.txt I'd sent out previously:
> 
> ---
> 15. Legacy
> ----------
> * cross-check that the 'pkg' attribute points to a package that depends on us
> * check that all the required tags are present on a legacy action and the 
> values
>   meet any SVR4 package restrictions
> ---
> 
> the first item being the one this lint check is reporting. For example,
> 
> pkg://opensolaris.org/text/[email protected],5.11-0.145:20100730T023438Z
> 
> delivers:
> 
> legacy arch=i386 category=system desc="gawk - GNU awk 3.1.5" \
>     hotline="Please contact your local service provider" \
>     name="gawk - GNU implementation of awk" pkg=SUNWgawk \
>     variant.arch=i386 vendor="Sun Microsystems, Inc." \
>     version=11.11.0,REV=2010.07.20.04.23
> 
> containing a pkg=SUNWgawk attribute. Here, we're checking that we can
> find a pkg://opensolaris.org/SUNWgawk package, and are ensuring it has a
> dependency on text/gawk, which it does, since it delivers:
> 
> depend fmri=text/[email protected] type=require
> 
> Given the usr/root package combination exercise we went through, this
> check is failing for quite a few packages.
> 
> I could jump through a few hoops and see if dropping a trailing 'u' or
> 'r' on the old package name yields any package that match, but that's
> likely to be inaccurate.  I suspect the best thing to do is to ignore
> missing ancestor packages like these, only reporting errors if we do
> find the ancestor package, and it hasn't got the correct dependency.

Yeah, since it wasn't only the root & usr that got combined, but sometimes
other combinations, like the CDE ones combined into what's now cde-runtime.

-- 
        -Alan Coopersmith-        [email protected]
         Oracle Solaris Platform Engineering: X Window System

_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to