May 2007 - macports-dev - lists.macosforge.org

Am I opening tickets improperly
by Glen Whitney 18 May '07

18 May '07

Dear MacPorts developers, I just wanted to make sure I've been opening tickets properly, since a few have not been getting feedback. That could easily be because you are busy with other aspects of MacPorts, not to mention the rest of your lives, but I thought I should just make sure. All of my tickets (except 1 against trac) are against the base, and so following the TracTicketing page on the wiki, I've labeled them with version numbers like base-1.4.40, just me in cc: (since "port info base" does not refer to the base, there was no email address to use for maintainer), and the base component. Are there folks that scan all the recent tickets against base? I ask because one of the bugs seems quite serious to me, Ticket #11920 (https://svn.macosforge.org/projects/macports/ticket/11920) This bug can cause a port built with the wrong variants turned on to be silently installed. An occurrence of this caused me significant consternation. It was pretty hard to track down, since it was installing the image _called_ emacs_devel 21.0.97+x11, and it took a long time to occur to me that that image had been compiled with -x11 (!). Anyhow, even though the symptoms can be pretty serious, the fix turns out not to be hard. But I have had the ticket with patch posted for a week without any comment. I tried to guess at the highest priority, I chose "Blocker", but apparently I guessed wrong. I'd just like to spare someone else the confusion I went through, and so I'm wondering if there's anything else I can do to help resolve this issue. I also have 4 other tickets open against base, all of which I've submitted patches for. Ticket #11892 got marked "needs developer review" after I mentioned it on this list a week or so ago. Ticket #11898 just needs a 1-word documentation fix to be resolved, the man page was ambiguous as someone (pipping?) pointed out in the ticket but it doesn't seem to have been visited again to make the fix. Ticket #11891 I just mentioned in my previous note. And ticket #11897, like 11920, has gotten no response. It's not anywhere near as serious, but it also has a much more trivial fix. My main project is to get a set of ports together which will allow compiling gtk apps with the new non-X11 port of gtk purely on top of quartz, with GnuCash as the main objective (since I've switched to using the quartz GnuCash myself and don't want to have to update it by hand). That project is hard to work on with these outstanding issues. On the other hand, coding is very active on macports base, so I've been left updating the patches to the latest SVN every several days, which doesn't feel so productive. At the moment, all the patches are current up to r25303. Thanks for any advice on anything else I can do to help ease the resolution of these issues. I realize this is a volunteer, open-software effort and I've tried to provide evidence that I'm ready to do my appropriate bit and not be a free rider. Does base need a test harness to vet patches and avoid issues like the recent reinplace bug? Is that something I might be able to contribute that could help get patches in more quickly in the future? I'd offer to put in time as a ticket-resolver, but obviously I haven't yet built the reputation/trust that would lead to the granting of such a role. I do plan in the next couple of days to submit a patch for ticket #8221 that someone else opened, it's pretty clear how to do it now that #11891 is out of the way. Perhaps that will also help. Best regards, and my thanks to everyone for all of the work that has gone into MacPorts. Glen Whitney

3 3

Re: [25313] trunk/dports/shells/zsh-devel
by Vincent Lefevre 18 May '07

18 May '07

On 2007-05-18 06:27:58 -0700, source_changes(a)macosforge.org wrote: > +post-destroot { > + # This manpage only includes every zsh manpage; as manpages are gzipped in MacPorts, > + # includes fail, so this one is useless. > + delete ${destroot}${prefix}/share/man/man1/zshall.1 It would be better to fix the bug. FYI, the manpages are gzipped under Linux (Debian) too, and there's no problem with the zshall man page. But IMHO, it is probably better not to compress the manpages (that should be a way to solve the problem). Another problem with the gzipped man pages is that the man command sometimes freezes. And disk space on desktop machines are no longer a problem for things like man pages (if it is on some machines, then there should be an option[*] not to include the manpages). [*] e.g. a variant with a common name. Also note that deleting the zshall man page will give the user wrong information since "man zshall" will load the zshall man page that comes with Mac OS X. -- Vincent Lefèvre <vincent(a)vinc17.org> - Web: <http://www.vinc17.org/> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/> Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon)

1 0

Re: macports@macports.org
by Ryan Schmidt 18 May '07

18 May '07

On May 18, 2007, at 02:23, Juan Manuel Palacios wrote: > On May 18, 2007, at 12:52 AM, Ryan Schmidt wrote: > >> We have 8 ports maintained by "macports(a)macports.org". Should that >> be changed to "nomaintainer(a)macports.org"? > > Totally! (simply put, there is no macports(a)macpors.org, while the > former has proper handling server side) > > Thanks for giving the ports tree some love, Ryan! Regards,... This love has been committed in r25309: http://trac.macosforge.org/projects/macports/changeset/25309 :-)

1 0

Re: MacPorts v1.4.41 released for selfupdate
by Glen Whitney 18 May '07

18 May '07

James Berry writes: > - `port info` now opens the Portfile and displays up-to-date > information. > This allows information about variants to be displayed (eg. > dependencies). > Also allow `port info` to default to current directory if no > name is provided > (eridius r25255). > - Evaluate variants at dportopen instead of dportexec. This > ensures that > the portfile represents the proper information if queried > before targets > are executed. It also prevents variants from being executed > twice if multiple > targets are executed separately on the portfile (ticket > #11296, eridius r25033). These two changes make it very easy to fix the bug I reported in http://trac.macports.org/projects/macports/ticket/11891 (variants ignored in deps action). All that has to happen is that the code Mr. Ballard put into action_info for this purpose be shared by action_deps, and I have submitted a patch to ticket 11891 which does just that. Hopefully the patch will help get that ticket resolved. Thanks very much, Glen Whitney

1 0

macports@macports.org
by Ryan Schmidt 18 May '07

18 May '07

We have 8 ports maintained by "macports(a)macports.org". Should that be changed to "nomaintainer(a)macports.org"? $ grep macports@macports */*/Portfile devel/premake/Portfile:maintainers macports(a)macports.org perl/p5-class-returnvalue/Portfile:maintainers macports(a)macports.org perl/p5-file-flat/Portfile:maintainers macports(a)macports.org perl/p5-file-ncopy/Portfile:maintainers macports(a)macports.org perl/p5-file-remove/Portfile:maintainers macports(a)macports.org perl/p5-pod-tests/Portfile:maintainers macports(a)macports.org perl/p5-prefork/Portfile:maintainers macports(a)macports.org perl/p5-test-inline/Portfile:maintainers macports(a)macports.org $

1 0

selfupdate is broken
by Vincent Lefevre 18 May '07

18 May '07

It seems that selfupdate got broken: prunille:~> sudo port -d selfupdate DEBUG: Rebuilding the MacPorts base system if needed. Synchronizing from file:///Users/vinc17/wd/macosx/dports DEBUG: /opt/local/bin/svn update --non-interactive "/Users/vinc17/wd/macosx/dports" It shouldn't update *my* SVN working copy. It was previously updating the standard path (?) with rsync. I suppose that it takes the paths from /opt/local/etc/ports/sources.conf, which contains here: file:///Users/vinc17/wd/macosx/dports file:///Users/vinc17/software/dports rsync://rsync.macports.org/dpupdate/dports But I use this for port installation only. -- Vincent Lefèvre <vincent(a)vinc17.org> - Web: <http://www.vinc17.org/> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/> Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon)

6 14

Re: [25292] trunk/dports/python/py-gsl/Portfile
by Ryan Schmidt 17 May '07

17 May '07

On May 17, 2007, at 09:01, source_changes(a)macosforge.org wrote: > Revision: 25292 > http://trac.macosforge.org/projects/macports/changeset/25292 > Author: jochen(a)macports.org > Date: 2007-05-17 07:01:08 -0700 (Thu, 17 May 2007) > > Log Message: > ----------- > Update to upstream version 0.9.1 > > Provide variants for the different array-object packages (numarray, > Numeric, numpy) supported by pygsl. > > Take over maintainership. It would be much more easy to see what substantive changes you made if you committed purely whitespace changes separately from functional ones. > Modified Paths: > -------------- > trunk/dports/python/py-gsl/Portfile > > Modified: trunk/dports/python/py-gsl/Portfile > =================================================================== > --- trunk/dports/python/py-gsl/Portfile 2007-05-17 13:52:50 UTC > (rev 25291) > +++ trunk/dports/python/py-gsl/Portfile 2007-05-17 14:01:08 UTC > (rev 25292) > @@ -1,29 +1,46 @@ > # $Id$ > > -PortSystem 1.0 > -PortGroup python24 1.0 > +PortSystem 1.0 > +PortGroup python24 1.0 > > -name py-gsl > -version 0.3.2 > -categories python science > -platforms darwin > -maintainers nomaintainer(a)macports.org > -description Python interface to the GNU scientific library > -long_description Python interface to the GSL, the GNU scientfic \ > - library\; a lot of functions, random number generators \ > - are available. > +name py-gsl > +version 0.9.1 > +categories python science > +platforms darwin > +maintainers openmaintainer(a)macports.org > jochen(a)macports.org > +description Python interface to the GNU Scientific > Library > +long_description Python interface to the GSL, the GNU > Scientfic \ > + Library. Variants for the use of different \ > + array-objects supported by pygsl are > provided. > > -homepage http://pygsl.sourceforge.net/ > -master_sites sourceforge:pygsl > -distname pygsl-${version} > -checksums md5 dde5f5a23dba8e63818dbfadb12e9a09 > +homepage http://pygsl.sourceforge.net/ > +master_sites sourceforge:pygsl > +distname pygsl-${version} > +checksums md5 72323e678ae5473d4a7d9cfc04dcb568 > > -depends_lib-append port:gsl \ > - port:py-numeric > +build.env CFLAGS="-I${prefix}/include/gsl -I$ > {prefix}/include/python2.4 -I${prefix}/include" \ > + LDFLAGS=-L${prefix}/lib > > -build.env CFLAGS="-I${prefix}/include -I${prefix}/include/gsl"\ > - LDFLAGS=-L${prefix}/lib > +variant numarray conflicts numpy Numeric description "Use numarray > as array-object" { > + depends_lib-append port:gsl \ > + port:py-numarray > + build.args --array-object=numarray > +} > +variant Numeric conflicts numarray numpy description "Use Numeric > as array-object" { > + depends_lib-append port:gsl \ > + port:py-numeric > + build.args --array-object=Numeric > +} > +variant numpy conflicts numarray Numeric description "Use numpy as > array-object" { > + depends_lib-append port:gsl \ > + port:py-numpy > + build.args --array-object=numpy > +} > +if { ![variant_isset numarray] && ![variant_isset Numeric] && ! > [variant_isset numpy] } { > + default_variants +numpy > +} > > -post-destroot { > - xinstall -m 644 ${worksrcpath}/README ${destroot}${prefix}/share/ > doc/${name} > + > +post-destroot { > + xinstall -m 644 ${worksrcpath}/README ${destroot}${prefix}/ > share/doc/${name} > }

1 0

Perl port failing
by Blair Zajac 17 May '07

17 May '07

Was there a change recently to the reinplace command? It seems to break the Perl ports: $ port -v install p5-crypt-blowfish ... ... Fixing packlist /opt/local/var/db/dports/build/ _opt_local_var_db_dports_sources_rsync.rsync.darwinports.org_dpupdate_dp orts_perl_p5-crypt-blowfish/work/destroot/opt/local/lib/perl5/ vendor_perl/5.8.8/darwin-thread-multi-2level/auto/Crypt/ Blowfish/.packlist Error: Target com.apple.destroot returned: reinplace: unknown flag '- s|/opt/local/var/db/dports/build/ _opt_local_var_db_dports_sources_rsync.rsync.darwinports.org_dpupdate_dp orts_perl_p5-crypt-blowfish/work/destroot||' Warning: the following items did not execute (for p5-crypt-blowfish): com.apple.activate com.apple.destroot com.apple.archive com.apple.install Error: Status 1 encountered during processing. Here's the code in the Perl 5 group: post-destroot { foreach packlist [exec find ${destroot}${perl5.lib} - name .packlist] { ui_info "Fixing packlist ${packlist}" reinplace "s|${destroot}||" ${packlist} } } Regards, Blair -- Blair Zajac, Ph.D. <blair(a)orcaware.com> Subversion training, consulting and support http://www.orcaware.com/svn/

3 3

MacPorts v1.4.41 released for selfupdate
by James Berry 17 May '07

17 May '07

MacPorts v1.4.41 has been released for self update. To install this release you should be able to: sudo port selfupdate (it will take an hour or so for it to reach the rsync repository). James ChangeLog for this release: Release 1.4.41 (16-May-2007 tagged at r25266 by jberry): - `port info` now supports new --index flag for using old behaviour (eridius r25256). - `port info` now opens the Portfile and displays up-to-date information. This allows information about variants to be displayed (eg. dependencies). Also allow `port info` to default to current directory if no name is provided (eridius r25255). - Add -E flag to reinplace, does the same thing as in sed (eridius r25251). - Add support for [nosync] flag in sources.conf to prevent a source from syncing during a `port sync` or `port selfupdate` (eridius r25236). Example: file:///Users/landonf/misc/MacPorts/dports [nosync] - Updating a file: source via svn update during port sync now uses the uid/gid of the owner of the working copy (eridius r25142). - Make port(1) convert description and long_description from lists to strings before printing to stdout; this enables the proper printing of semi-colons, brackets and other characters by way of Tcl escape sequences. The exceptions to this are the sequences for newline and carriage return (these break the flatfile PortIndex format by introducing line breaks). (boeyms r25110) - Evaluate variants at dportopen instead of dportexec. This ensures that the portfile represents the proper information if queried before targets are executed. It also prevents variants from being executed twice if multiple targets are executed separately on the portfile (ticket #11296, eridius r25033). - Fix copy and move commands (eridius r25024). - Passing multiple arguments to $option-delete now works as expected (ticket #11929, eridius r25006). - fs-traverse no longer follows symlinks specified on the root. This fixes the incorrect behaviour of delete seen in #11862 (eridius r25003). - Remove bundled xar. (jberry r24913). - Tighten checking of port name in depspecs. neither + nor / is allowed. (jberry r24929, r24934, r24940). - Restore ability for portindex lines to contain carriage returns, allowing fields such as descriptions to contain line breaks. Note that any portindex containing such entries will not be readable by a recent version of macports. (jberry r25146) - Add support for descriptions on variants: * The syntax follows the syntax for variant depends, etc. * Descriptions for variants are entirely optional * The change to the portindex format to allow variant descriptions should be backwards compatible. Variant descriptions are stored, keyed by the variant name, in a new portinfo array named "variant_desc". Example: variant myvariant description "This variant does this and that" { configure.args-append --do-this --do-that } (jberry r25147) - Only look for, or create, ~/.macports user directory if the HOME environment variable is defined. (jberry r25177)

2 2

MacPorts v1.4.42 released for selfupdate
by James Berry 17 May '07

17 May '07

MacPorts v1.4.42 has been released for self update, correcting a bug in v1.4.41. To install this release you should be able to: sudo port selfupdate (it will take an hour or so for it to reach the rsync repository). James ChangeLog for this release: Release 1.4.42 (17-May-2007 tagged at r25296 by jberry) - Fix reinplace, broken in 1.4.41 (eridius r25285).

1 0