Date: Mon, 14 Apr 2003 11:48:51 -0700 From: Kent Stewart <kstewart@owt.com> To: Mike Meyer <mwm-dated-1050774277.76f83b@mired.org>, dick hoogendijk <dick@nagual.st> Cc: freebsd-questions <freebsd-questions@freebsd.org> Subject: Re: Error in fontconfig Message-ID: <200304141148.51008.kstewart@owt.com> In-Reply-To: <16026.62341.42973.753668@guru.mired.org> References: <5.2.0.9.0.20030414130808.00a457a0@127.0.0.1> <20030414170827.GA32192@lothlorien.nagual.st> <16026.62341.42973.753668@guru.mired.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday 14 April 2003 10:44 am, Mike Meyer wrote: > In <20030414170827.GA32192@lothlorien.nagual.st>, dick hoogendijk <dick@nagual.st> typed: > > On 14 Apr Kent Stewart wrote: > > > If you aren't rebuilding your INDEX* files after you cvsup, you > > > are using an INDEX dated 29 Mar. > > > > Is it really needed to rebuild INDEX after every cvsup of the > > ports? Someone suggested one time: > > # make index > > # pkgdb -F > > # portsdb -u > > I think the "pkgdb -F" isn't needed. The database of installed > packages isn't really affected that much by cvsupping ports. About > the only thing that might break is that a port may have been renamed > or obsoleted, so that the db no longer has a valid origin. That won't > make any difference until you try and upgrade that port, at which > point you'll be told to run "pkgdb -F". > > The pkgdb man page says run it "periodically". That could be when you > do a ports cvsup. Personally, I run it when one of the portupgrade > tools says it's broken and I need to run it. > > > This all takes a long time. "Make index" takes 'ages' > > I guess I have to live with that if it means you're errorfree, but > > still, is it? > > Yes, it is. Note that "portsdb -U" will do the equivalent of "make > index" for you. So all you have to do is one "portsdb -uU" after > cvsupping the ports. Which is exactly what the portsdb man page > recommends doing. There have been problems in the recent past when you used -U. On 4.x, you would get one type of breakage with "make index" and a different one with portsdb -U. With portsdb -U, you currently get <100's of messages snipped> make_index: ja-balsa-0.8.0: no entry for /usr/local Warning: Duplicate INDEX entry: freeciv-imlib-1.13.0 done 327.930u 180.856s 10:58.71 77.2% 260+491k 49418+34271io 77pf+0w ruby# portsdb -u [Updating the portsdb <format:bdb1_btree> in /usr/ports ... - 8479 port entries found .........1000.........2000.........3000.../usr/ports/INDEX:3343:darkbot-6f6_r6: 6f6_r6: Not in due form: '<version>[_<revision>][,<epoch>]'. ......4000.........5000.........6000.........7000.........8000.... ..... done] With make index, you get ruby# time make index Generating INDEX - please wait.. Done. 453.173u 141.729s 17:30.97 56.6% 247+5794k 41174+110io 80pf+0w ruby# portsdb -u [Updating the portsdb <format:bdb1_btree> in /usr/ports ... - 8490 port entries found .........1000.........2000.........3000.../usr/ports/INDEX:3347:darkbot-6f6_r6: 6f6_r6: Not in due form: '<version>[_<revision>][,<epoch>]'. ......4000.........5000.........6000.........7000.........8000.... ..... done] You currently don't see all of the messages with make index and it produces a list with 11 more ports. There is a cost because make index takes almost 7 minutes of wall clock time longer to run on an AMD 2000+. Kent -- Kent Stewart Richland, WA http://users.owt.com/kstewart/index.html
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200304141148.51008.kstewart>