Date: Sun, 17 Oct 2004 12:59:19 +0300 From: Claudiu Costin <claudiuc@kde.org> To: freebsd-cvsweb@freebsd.org Subject: Re: cvsweb will not work like before :( Message-ID: <200410171259.20259@kde_is_very_cool> In-Reply-To: <1097926338.2719.23.camel@bobcat.mine.nu> References: <200410142354.43039@kde_is_very_cool> <58670.69.53.57.66.1097788341.squirrel@69.53.57.66> <1097926338.2719.23.camel@bobcat.mine.nu>
next in thread | previous in thread | raw e-mail | index | archive | help
Dear Ville,
Thank you very much. I can't wait to see the feature working
again. Please give me a sign (a patch) when it's done. I'm
subscribed on the list.
On S=EEmb=E3t=E3 16 Octombrie 2004 14:32, Ville Skytt=E4 wrote:
> On Fri, 2004-10-15 at 00:12, Jon Noack wrote:
>=20
> > > For example in my PHP script I have the following printf() like string
> > > variable:
> > > $cvswebformat2=3D"http://webcvs.kde.org/cgi-bin/cvsweb.cgi/~checkou=
t~/kde-i18n/%s?rev=3D%s&content-type=3Dtext/plain";
> > > where first "%s" is like fr/kdeadmin/kcron.po, and second "%s" is bra=
nch
> > > like KDE_3_3_BRANCH or HEAD.
> >=20
> > Actually, it works fine for "HEAD" or ".". According to the Changelog,
> > this functionality was "resurrected" on 2004-04-20. However, other
> > branches/tags I tried did not work.
>=20
> Right. It "works" for HEAD, because the requested tag/branch seems to
> be ignored altogether, and the magic revision "." currently results in
> -rHEAD being passed to "cvs co".
>=20
> I already started enhancing the "." revision so that it takes the branch
> into account, it seems to work pretty ok here. But there's a catch:
> https://ccvs.cvshome.org/issues/show_bug.cgi?id=3D211
>=20
> What the above means is that before one can do a "cvs co -p" using a
> branch name (ie. -rBRANCH, =3D=3D "latest on branch BRANCH"), BRANCH needs
> to be in CVSROOT/val-tags. "cvs co -p" does not put it there, but "cvs
> co" (without -p) does -> someone has to do the latter outside of cvsweb
> before it starts to work in it :( Additionally, the error messages from
> "cvs co -p" for nonexistent tags/branches are, well, non-obvious. This
> is as of cvs 1.11.17.
>=20
> I think I'll commit this stuff anyway today. If the above bug 211
> doesn't get fixed in cvs soonish and this starts to cause problems, I'll
> take a look at changing "cvs co" to checkout into a temporary file in
> cvsweb, then operating on that instead of directly using the STDOUT from
> "cvs co -p". That might not be a bad idea in the first place, BTW.
>=20
>=20
kind regards,
=2D-=20
Claudiu Costin, claudiuc@kde.org
Linux-KDE Romania http://www.ro.kde.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200410171259.20259>