On Sun, 02 Jan 2011 16:03:41 0100, Martin Costabel wrote:
On 2/01/11 09:33, Daniel Macks wrote:
> > (copying back to to list...already sent to maintainer and poster)
> >
> > On Thu, 30 Dec 2010 13:09:20 0100, Antoine Pérus wrote:
> >
> > Warning: Incorrect filename 'visual-py26.info'. Should be one of:
> > visual-py.info
> > visual-py-5.32.info
> > visual-py-5.32-1.info
>
> This one I won't fix. I dispute the wisdom of the validator to impose
> such rules.
A parenthesized list of Type is only for multivalue variants. If you
just have one, do not parenthesize it. This:
Type: python (2.6)
is a package that is written for multiple versions but only happens to
use one at this time and would be in visual-py.info, whereas:
Type: python 2.6
is a pure python2.6 package that would be fine as visual-py26.info. The
current situation mixes those two ideas? Where's the sanity in putting
a py27 variant in a -py26.info file and leaving the portable-looking
"-py.info" file as a dead-end old version up to py25?
> > Error: PatchFile-MD5 does not match PatchFile checksum.
> (visual-py26.info)
> > Actual: 99ffc72f955893b103dece7e5e6162ed
> > Expected: caf39936d19921bea86d10600cc2cf78
>
> This is another bug in the validator. In fact, if the md5sum were really
> wrong, the package would fail to build even without the -m flag.
> Fink takes visual-py26.patch as patchfile (as it should), whereas the
> validator takes visual-py.patch. In the info file, it is %n.patch. Since
> the new version allows py27, too, I have now hardwired the patchfile
> name to visual-py26.patch, so this validator bug does no longer show up.
Now you have:
Package: visual-py%type_pkg[python]
Type: python (2.6 2.7)
PatchFile: %n.patch
which fails to build (even without -m flag) for the 2.7 variant:
Failed: Cannot read PatchFile
"/sw/fink/dists/unstable/main/finkinfo/sci/visual-py27.patch"
Sure enough, it doesn't exist. Again, just using the standard
visual-py.{info,patch} filenaming (and PatchFile:%{ni}.patch standard
incantation) would solve all these messes cleanly.
dan
--
Daniel Macks
[email protected]
------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and,
should the need arise, upgrade to a full multi-node Oracle RAC database
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl
_______________________________________________
Fink-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/fink-users
