[clc-devel] [CRUX] #67: kdepim unbuildable with --enable-final on low mem systems
#67: kdepim unbuildable with --enable-final on low mem systems ---------------------+------------------------------------------------------ Id: 67 | Status: new Component: ports | Modified: Sat Jan 28 16:17:24 2006 Severity: trivial | Milestone: Priority: normal | Version: Owner: danm | Reporter: mark ---------------------+------------------------------------------------------ I tried building kdepim on a system with 128M RAM and 256M swap. The result was that kswapd0 freaked out, the CPU usage was 100% of which 1% were used by g++. In order to build, I disabled --enable-final. Nothing but the default rc procs were running. I'm all for having this optimization enabled in the Pkgbuild, but I suggest mentioning this issue in the README. kdepim is the only port that needed this special treatment; the other kde* ports I've tried builds fine with --enable-final. -- Ticket URL: <http://crux.nu/cgi-bin/trac.cgi/ticket/67> CRUX <http://crux.nu/> CRUX
#67: kdepim unbuildable with --enable-final on low mem systems ---------------------+------------------------------------------------------ Id: 67 | Status: new Component: ports | Modified: Sat Jan 28 23:44:09 2006 Severity: trivial | Milestone: Priority: normal | Version: Owner: sten | Reporter: mark ---------------------+------------------------------------------------------ Changes (by sten): * owner: danm => sten Comment: Mark, could you please try running the following command, before re- building kdepim (after a ports -u opt, of course): sed -i 's/make$/make no-final/' /usr/ports/opt/kdepim/Pkgfile According to (http://developer.kde.org/documentation/other/developer- faq.html), `"make no-final" does a normal compilation in the current directory even if --enable-final was used"'. Depending on how many optional dependencies a low-mem user has installed, it might even be necessary to: sed -i 's/make$/make no-final/' /usr/ports/opt/kde*/Pkgfile At any rate, please test this, and tell me how it goes. -- Ticket URL: <http://crux.nu/cgi-bin/trac.cgi/ticket/67> CRUX <http://crux.nu/> CRUX
#67: kdepim unbuildable with --enable-final on low mem systems ---------------------+------------------------------------------------------ Id: 67 | Status: new Component: ports | Modified: Sun Jan 29 03:19:47 2006 Severity: trivial | Milestone: Priority: normal | Version: Owner: sten | Reporter: mark ---------------------+------------------------------------------------------ Comment (by mark): How is "make no-final" different from not specifying --enable-final? (Which is what I did to make it build, I just thought it should be mentioned in the README because 128M isn't *that* uncommon yet :-)) -- Ticket URL: <http://crux.nu/cgi-bin/trac.cgi/ticket/67> CRUX <http://crux.nu/> CRUX
#67: kdepim unbuildable with --enable-final on low mem systems ---------------------+------------------------------------------------------ Id: 67 | Status: assigned Component: ports | Modified: Sun Jan 29 05:30:05 2006 Severity: trivial | Milestone: Priority: normal | Version: Owner: sten | Reporter: mark ---------------------+------------------------------------------------------ Changes (by sten): * status: new => assigned Comment: "make no-final" requires editing one line, as opposed to two. Also, the sed command is legible to the point of being almost self-evident, so I feel that users would understand it, trust it, and copy & paste it immediately. Yes, I will also make a note that it can be done manually :-). I don't have a low-mem system to test whether or not it's different from deleting the --enable-final line, and removing the previous line's "\", so would you please try out the copy & paste solution? Furthermore, one wouldn't have to edit one's /usr/ports/opt/kde*/Pkgfile after every ports -u, if one added that sed command to the end of it. Personally, I like CRUX because it can be customized to get out of way, and do things in a sane fashion; editing a Pkgfile after every ports -u seems awfully labour intensive to me. ;-) -- Ticket URL: <https://crux.nu/cgi-bin/trac.cgi/ticket/67> CRUX <http://crux.nu/> CRUX
#67: kdepim unbuildable with --enable-final on low mem systems ---------------------+------------------------------------------------------ Id: 67 | Status: assigned Component: ports | Modified: Sun Jan 29 06:30:46 2006 Severity: trivial | Milestone: Priority: normal | Version: Owner: sten | Reporter: mark ---------------------+------------------------------------------------------ Comment (by mark): I'll try when I get to the box again. Personally I use a "real" SVN checkout of the repo kept in ~/src - that way I have a good overview of what ports I've changed, and unless the committer happens to change the same line as me (hasn't happened yet), even merging happens automagically :-) -- Ticket URL: <http://crux.nu/cgi-bin/trac.cgi/ticket/67> CRUX <http://crux.nu/> CRUX
#67: kdepim unbuildable with --enable-final on low mem systems ---------------------+------------------------------------------------------ Id: 67 | Status: assigned Component: ports | Modified: Fri Feb 3 13:28:14 2006 Severity: trivial | Milestone: Priority: normal | Version: Owner: sten | Reporter: mark ---------------------+------------------------------------------------------ Comment (by mark): Sorry, I don't have the box available any more :-/ It does look like make no-final should do the same thing, though. As for the other kde* ports, kdepim was the only one that had the problem on this system. IIRC it was the kmail build that triggered it. -- Ticket URL: <http://crux.nu/cgi-bin/trac.cgi/ticket/67> CRUX <http://crux.nu/> CRUX
#67: kdepim unbuildable with --enable-final on low mem systems ---------------------+------------------------------------------------------ Id: 67 | Status: closed Component: ports | Modified: Fri Feb 3 18:59:36 2006 Severity: trivial | Milestone: Priority: normal | Version: Owner: sten | Reporter: mark ---------------------+------------------------------------------------------ Changes (by sten): * resolution: => fixed * status: assigned => closed Comment: No problem. I guess I'll say that this was fixed in [649] . -- Ticket URL: <http://crux.nu/cgi-bin/trac.cgi/ticket/67> CRUX <http://crux.nu/> CRUX
participants (1)
-
CRUX