On Sat, Sep 18, 2004 at 10:48:33PM -0600, Michael G. Goodell wrote: > When I did a make "install clean" on the /usr/ports/x11/kde3 port it built > for a quite a while then died with this error: > > /usr/local/include/kjs/lookup.h:96: warning: non-static const member `const > char* const KJS::HashTable::sbase' in class without a constructor gmake[3]: > *** [SVGColorProfileElement.lo] Error 1 gmake[3]: Leaving directory > `/usr/ports/graphics/kdegraphics3/work/kdegraphics-3.3.0/ksvg/dom' gmake[2]: > *** [all-recursive] Error 1 gmake[2]: Leaving directory > `/usr/ports/graphics/kdegraphics3/work/kdegraphics-3.3.0/ksvg' gmake[1]: *** > [all-recursive] Error 1 gmake[1]: Leaving directory > `/usr/ports/graphics/kdegraphics3/work/kdegraphics-3.3.0' gmake: *** [all] > Error 2 *** Error code 2 Stop in /usr/ports/graphics/kdegraphics3. *** > Error code 1 Stop in /usr/ports/x11/kde3. > > I was told that I should replace the file /usr/include/sys/endian.h with a > new one that was emailed to me by a maintainer. It seems to have worked ok > but my question is this: I did a complete cvsup update just prior to > building kde3 and built he kernel and world on: Aug 27 22:24:12 MDT 2004 - > would not have rebuilding my kernel / cvsup fetched this new endian.h file? > If not why?
That's the thing when corresponding with developers: they may well ask
you to test out patches they've literally just written. Seeing as you
seem to have had a good result from that, I'd expect the patch (or
something equivalent) to make it's way into the system eventually.
You didn't say if you were corresponding with a KDE developer or a
FreeBSD kernel developer -- as the file that was replaced is part of
the FreeBSD system, a KDE person would have to pass it over to the
FreeBSD side for review.
Even so, the patch would generally be committed to 6-CURRENT first,
for testing, and then merged into 5-STABLE and maybe 4-STABLE after
some time. At the moment however 5-STABLE is in a source code freeze
prior to the release of 5.3-RELEASE, so anything added there has to
pass through the even more rigourous examination of the release
engineering team.
On the other hand, it could be a well-known fix that enables that
software to build on an older version of the system: you don't say
which system version you're running, but if it's older than
4.10-RELEASE or it it's one of the 5.x preview releases (5.2.1-RELEASE
or earlier) -- then that patch wouldn't be applied to those versions[1].
In which case, you'ld either have to always apply the patch manually
whenever you did a buildworld, or you'ld have to upgrade to a newer
version of the system where it would be incorporated.
Or the KDE folks might possibly incorporate a work-around into the kde3
port to enable it to build cleanly on earlier system versions.
> I hope I posted this OK - I am new to the mailing list thing.
Seems fine to me.
Cheers,
Matthew
[1] The project guarrantees that the -RELEASE branches will maintain a
stable API/ABI and that only security patches will be applied.
Although they have moved to saying 'Security+Major Errata' for
4.10-RELEASE.
--
Dr Matthew J Seaman MA, D.Phil. 26 The Paddocks
Savill Way
PGP: http://www.infracaninophile.co.uk/pgpkey Marlow
Tel: +44 1628 476614 Bucks., SL7 1TH UK
pgpoN5vCIPKhZ.pgp
Description: PGP signature
