Date: Sat, 29 Sep 2007 06:35:25 +0200 (CEST) From: ports-x11-manpages-may07@oldach.net (Helge Oldach) To: edwin@FreeBSD.org (Edwin Groothuis) Cc: freebsd-x11@FreeBSD.org Subject: Re: ports/113096: [patch] x11/xorg - X.org manpages broken Message-ID: <200709290435.l8T4ZPCd020687@sep.oldach.net> In-Reply-To: <200709280244.l8S2iFw5064159@freefall.freebsd.org> from "edwin@FreeBSD.org" at "Sep 28, 2007 2:44:15 am"
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Edwin, edwin@FreeBSD.org: > Synopsis: [patch] x11/xorg - X.org manpages broken > > State-Changed-From-To: open->closed > State-Changed-By: edwin > State-Changed-When: Fri Sep 28 02:32:57 UTC 2007 > State-Changed-Why: > This is fixed in the man(1) command in HEAD, RELENG_5 and RELENG_6. > It will be available in 7.0, 6.3 and 5.6 or when you apply the patch at > http://wwwfe.freebsd.org/cgi/cvsweb.cgi/src/gnu/usr.bin/man/man/man.c.diff?r1=1.63;r2=1.65 > > http://www.freebsd.org/cgi/query-pr.cgi?pr=113096 > Not quite. Indeed you've fixed this for the man(1) case. However, as you know one can also preformat manpages by means of catman(1), for instance by setting weekly_catman_enable="YES" in periodic.conf. But catman(1) still does it wrong and generates a blank manpage for source files that contain a ".so something", which man(1) later simply displays instead of using the .so directive. In other words, catman(1) should be fixed with a similar patch to really fix this bug. Furthermore, the MLINKS feature creates a set of soft-linked manpages which catman(1) correctly converts to a *single* catpage file that is hard-linked, for example: -rw-r--r-- 2 man wheel 332 Sep 15 22:16 cat3/XtProcessLock.3.gz -rw-r--r-- 2 man wheel 332 Sep 15 22:16 cat3/XtProcessUnlock.3.gz -r--r--r-- 1 root wheel 1680 Sep 15 18:10 man3/XtProcessLock.3.gz lrwxr-xr-x 1 root wheel 18 Sep 15 18:10 man3/XtProcessUnlock.3.gz@ -> XtProcessLock.3.gz Note that in some cases (particularly with xorg) we have some 20 or 30 soft-linked manpages, which implies significant savings in terms of disk space and CPU if they all point to a single hard-linked catpage instead of having an extra file for each manpage with identical contents. So it appears that the appropriate catman(1) fix that would be slightly more difficult to emulate the MLINKS behaviour completely by means of .so directives... MLINKS is not obsolete yet. Note there is also bin/96540 that fixes a corner case for hard-linked catpages with catman(1). Regards, Helge
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200709290435.l8T4ZPCd020687>