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

Re: [28702] trunk/base/HACKING
by Jochen Küpper 19 Sep '07

19 Sep '07

On 07.09.2007, at 20:34, source_changes(a)macosforge.org wrote: > + # -*- Mode: Tcl; tab-width: 4; indent-tabs-mode: nil; c- > basic-offset: 4 -*- vim:et:sw=4:ts=4:sts=4 Shouldn't there be a "coding: utf-8" in there, at least for Portfiles? Greetings, Jochen -- Einigkeit und Recht und Freiheit http://www.Jochen- Kuepper.de Liberté, Égalité, Fraternité GnuPG key: CC1B0B4D Sex, drugs and rock-n-roll

5 5

HDF 5 fails to upgrad
by Michael Wild 17 Sep '07

17 Sep '07

Hi all I submitted a bug concerning HDF-5 which fails to upgrade because the test programs link against the old (still active) version. http://trac.macports.org/projects/macports/ticket/12705 Regards Michael

1 0

Dynamic libraries, gtk-sharp
by bob spamnot 16 Sep '07

16 Sep '07

I am trying to build gtk-sharp-2.0 from source since there is no working port of this (and there isn't on Fink either at the moment). If I can get this to work I can make a portfile for it, so I guess it does relate to Macports:-). But I cannot get the dynamic library situation sorted. In fact I am quite mystified by the apple DYLN_LIBRARY_PATH variable. Any help or pointers will be greatly appreciated. Sorry about the long post, I've tried to condense it down. Here is the steps I have done (note I am on a PowerBook G4, OS X 10.4.10, XCode 2.4.1): 1a. Install gnome: sudo port install gnome Actually I haven't (yet) got it to fully install but it makes it most of the way (certainly far enough for the bits needed by gtk-sharp-2.0). 1b. Install mono: sudo port install mono 2. Download gtk-sharp from gnome snv to a local src directory. I've tried both the HEAD version and tagged version 2.10.2. 3. Make it in the following way: bash> echo $PATH usr/local/bin:/opt/local/bin:/opt/local/sbin:/usr/X11R6/bin:/bin:/sbin:/usr/bin:/usr/sbin bash> ./configure --prefix=/Users/name/local-sharp bash> make bash> make install This all works, so now gtk-sharp is install in local-sharp in my home directory. 4. Now I try to compile a simple program: bash> which mcs mcs is hashed (/opt/local/bin/mcs) bash> export PKG_CONFIG_PATH=/opt/local/lib/pkgconfig:/usr/X11R6/lib/pkgconfig:/Users/name/local-sharp/lib/pkgconfig bash> mcs -pkg:gtk-sharp-2.0 HelloWorldGtk.cs bash> mono HelloWorldGtk.exe ** (HelloWorldGtk.exe:10680): WARNING **: The following assembly referenced from /Users/name/src/gtk-sharp-cvs/test/HelloWorldGtk.exe could not be loaded: Assembly: gtk-sharp (assemblyref_index=1) Version: 2.10.0.0 Public Key: 35e10195dab3c99f System error: Unknown error: 0 ** (HelloWorldGtk.exe:10680): WARNING **: Could not load file or assembly 'gtk-sharp, Version=2.10.0.0, Culture=neutral, PublicKeyToken=35e10195dab3c99f' or one of its dependencies. Unhandled Exception: System.IO.FileNotFoundException: Could not load file or assembly 'gtk-sharp, Version=2.10.0.0, Culture=neutral, PublicKeyToken=35e10195dab3c99f' or one of its dependencies. File name: "gtk-sharp, Version=2.10.0.0, Culture=neutral, PublicKeyToken=35e10195dab3c99f" OK, I guess I need to set DYLD_LIBRARY_PATH bash> export DYLD_LIBRARY_PATH=/opt/local/lib:/Users/name/local-sharp/lib (note leaving out /opt/local/lib doesn't work either) bash> mono HelloWorldGtk.exe dyld: Symbol not found: __cg_jpeg_resync_to_restart Referenced from: /System/Library/Frameworks/ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/Versions/A/ImageIO Expected in: /opt/local/lib/libJPEG.dylib Trace/BPT trap Here is where I'm stuck. I've tried lots of variations but they all either give me this error or they cannot find gtk-sharp-2.0 or the regular gtk libs (which are in /opt/local/lib). I'll spare you the details of the other variations except one. If I set DYLD_LIBRARY_PATH=/opt/local/lib and just try to use eog or mcs I get this same error (note they both work fine if DYLN is not set). E.G.: bash> export DYLD_LIBRARY_PATH=/opt/local/lib bash> eog dyld: Symbol not found: __cg_jpeg_resync_to_restart Referenced from: /System/Library/Frameworks/ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/Versions/A/ImageIO Expected in: /opt/local/lib/libjpeg.62.dylib Trace/BPT trap bash> mcs HelloWorld.cs (very simple helloworld which just writes to console and works if DYLN not set) dyld: Symbol not found: __cg_jpeg_resync_to_restart Referenced from: /System/Library/Frameworks/ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/Versions/A/ImageIO Expected in: /opt/local/lib/libJPEG.dylib Trace/BPT trap This seems completely bizarre to me. Both eog and mcs were made by Macports and linked against stuff in /opt/local/lib, so why should adding /opt/local/lib to DYLD_LIBRARY_PATH make libJPEG look into the framework listed above. Maybe someone could explain to me what DYLN_LIBRARY_PATH is doing (the man page doesn't really help)? I've try searching google on the above "Symbol not found" error and get lots of hits (so I'm not alone) but none of the suggestions seem to help. I'd be glad to show some of the other things I've tried but this post is long enough and I'm hoping someone will have an easy answer:-) I hope I was clear enough, feel free to ask for clarification. Thanks, Bob

2 1

"Documentation" milestone renamed
by Juan Manuel Palacios 16 Sep '07

16 Sep '07

Evening everyone! I recently renamed the "Documentation" milestone up on our roadmap to "Website & Documentation", so that tickets for the trunk/www space that's currently being rewritten can be grouped there too (tickets have the option of a 'www' component, which should be a natural choice). Unfortunately, the new name tosses the milestone to the end of the roadmap, so feel free to suggest a new/better one if you have any ideas. On a related note, the new website is shaping up pretty good, albeit a bit slowly as Chris and I haven't had much time lately. In any case, I do believe it is in a stage where it can already receive criticism, so please do voice your ideas and opinions (either on this list or as trac tickets, or both) if you have any input on the content and/or structure. As always, trunk/www HEAD is always being made available at my server: http://apollo.homeunix.net/macports And my ideas for how the end product should look like are contained at the end of a message on the topic I posted in August: http://lists.macosforge.org/pipermail/macports-dev/2007-August/ 002642.html I do welcome any feedback on that and/or what's currently in trunk/ www. For the record, next up on my list is the rewrite of the main index.php page and of the last two sections ("Help" & "SVN Sources") of getmp.php. Regards,... -jmpp

3 2

Proposed change to "Target" term in base ui_msg's
by markd＠macports.org 16 Sep '07

16 Sep '07

Proposal: Changing the ui_msgs in porttrace.tcl and portutil.tcl. from: "Target $target returned/has a blah blah" to "MacPorts Installation Phase $target returned/has a ...." or just "Installation Phase $target returned/has a ....." Reasoning: I think the term "target" in MacPorts user-level communication is ambiguous and confusing. It invites the question, "Target of what?" At the level of user-level messages at least, it seems "target" is more or less synonymous with "phase", since "target phase" or just "phase" seems to communicate the same thing more clearly. But "phase" alone seems inadequate because that term invites the question "What type of phase?" so I'd prefer the term "install phase", assuming that it is ok to call all the MP phases generically "install phases", even though technically install is a phase itself. But that seems much better than the way the old guide speaks of "build targets" generically, because it seems more natural to call the MP phases "install phases" because "building" has a more specific meaning and makes "build phases" seem innaccurate (or at least confusing) as a term for MP phases. What do you all think about this? I want to harmonize the error messages with the new guide and I've already expunged the term "target" for the reasons stated above. Here are the ui_msgs as they are now. --------------- port1.0/porttrace.tcl: ui_warn "Target $target has an undeclared dependency on $port" port1.0/porttrace.tcl: ui_debug "Target $target has no traceable dependency on $port" port1.0/portutil.tcl: ui_error "Target $name returned: $errstr" port1.0/portutil.tcl: ui_debug "$ident registered provides '$target', a pre-existing procedure. Target override will not be provided" --------------- Mark

2 2

[12591] pandoc Portfile update
by John MacFarlane 16 Sep '07

16 Sep '07

Hello all, I've updated the Portfile for textproc/pandoc to build version 0.44. I've also modified it to install the Haskell library and library documentation in addition to the program and user documentation. It would be great if someone with commit rights could take a look at the new Portfile and (if it looks okay) commit it: http://trac.macosforge.org/projects/macports/ticket/12591 Many thanks, John

1 0

[doxygen] dependency on teTeX
by Mij 15 Sep '07

15 Sep '07

hello teTeX has been discontinued and the people is moving away from it, typically to texLive. Is it possible to remove the dependency of macport's doxygen to teTeX, so that you don't have to keep two huge versions of latex installed?

2 1

gnome build issues
by bob spamnot 15 Sep '07

15 Sep '07

I'm trying to switch from Fink to Macports cause it has much newer ports (thanks for that) and am trying to get gnome to install on a fresh install of Macports 1.52 (note that I've moved fink out of the way). Here are some of the issues I've run into (note that the first two I've found workarounds for but the 3rd is a stopper). I'm on a PowerBook G4, OS X 10.4.10, XCode 2.4.1. 1. esound issues: esound wouldn't install cause the configure script just ends w/o doing anything. There have been several other people having the same problem in the last few days on the Macports-Users board. I discovered that making the esound port skip the autoconf allows it to install (i.e., I did sudo port edit esound and switched the use_autoconf line from yes to no). I'm not sure if this is the correct way to fix it but if it is it should corrected at the Macports end. 2. gnome-applets issues: the gnome-applets configure script dies trying to satisfy python >= 2.4. I had python2.5 installed (by some port) but this was not good enough. So I installed python2.4 and gnome-applets would install. If this is correct then python2.4 should be added as a dependency. Otherwise the configure script should be fixed to detect python2.5 correctly. 3. seahorse: This is presently a stopper for me. Doing: bash$ sudo port clean seahorse ---> Cleaning seahorse sudo port clean seahorse bash$ sudo port -dv install seahorse DEBUG: Found port in file:///opt/local/var/macports/sources/rsync.macports.org/release/ports/gno… DEBUG: Changing to port directory: /opt/local/var/macports/sources/rsync.macports.org/release/ports/gnome/seahorse LOTS OF LINES DELETED /bin/sh ../../libtool --tag=CC --mode=link gcc -O2 -L/opt/local/lib -lpth -o seahorse-preferences seahorse-pgp-preferences.o../../libseahorse/libseahorse.a ../../libcryptui/libcryptui.la -Wl,-framework,CoreAudio -L/opt/local/lib -L/usr/X11R6/lib -lgnomeui-2 -lbonoboui-2 -lgnome-keyring -lgnomecanvas-2 -lgnome-2 -lpopt -lart_lgpl_2 - lpangoft2-1.0 -lesd -lbonobo-2 -laudiofile -lbonobo-activation -lglade-2.0-lxml2 -lpthread - lgtk-x11-2.0 -lgdk-x11-2.0 -latk-1.0 -lgdk_pixbuf-2.0 -ltiff -ljpeg - lpangocairo-1.0 -lpango-1.0 -lcairo -lSM -lICE -lfreetype -lz -lfontconfig -lexpat -lpng12 -lXrender -lX11 -lgnomevfs-2 -lgconf-2 -lORBit-2 -lm - lgmodule-2.0 -lgthread-2.0 -lgobject-2.0 -lglib-2.0 -lintl -liconv -L/opt/local/lib -lgpgme -L/opt/local/lib -lgpg-error -lldap -L/opt/local/lib -lsoup-2.2 -lglib-2.0 -lintl -lxml2 -lpthread -lz -liconv -lm -L/opt/local/lib -lgnome-keyring -lglib-2.0 -lintl -liconv -L/opt/local/lib -lavahi-common -lavahi-client -lavahi-glib -lglib-2.0-lintl -liconv -L/opt/local/lib -ldbus-glib-1 -ldbus-1 - lgobject-2.0 -lglib-2.0 -lintl -liconv mkdir .libs gcc -O2 -o .libs/seahorse-preferences seahorse-pgp-preferences.o-Wl,-framework -Wl,CoreAudio -L/opt/local/lib ../../libseahorse/libseahorse.a ../../libcryptui/.libs/libcryptui.dylib /opt/local/lib/libpth.dylib -L/usr/X11R6/lib /opt/local/lib/libgnomeui- 2.dylib /opt/local/lib/libbonoboui-2.dylib /opt/local/lib/libgnomecanvas- 2.dylib /opt/local/lib/libgnome-2.dylib /opt/local/lib/libpopt.dylib /opt/local/lib/libart_lgpl_2.dylib /opt/local/lib/libesd.dylib /opt/local/lib/libbonobo-2.dylib /opt/local/lib/libaudiofile.dylib /opt/local/lib/libbonobo-activation.dylib /opt/local/lib/libORBitCosNaming- 2.dylib /opt/local/lib/libglade-2.0.dylib /opt/local/lib/libgtk- x11-2.0.dylib /opt/local/lib/libgdk-x11-2.0.dylib /opt/local/lib/libatk- 1.0.dylib /opt/local/lib/libgdk_pixbuf-2.0.dylib/opt/local/lib/libtiff.dylib /opt/local/lib/libjpeg.dylib /opt/local/lib/libpangocairo-1.0.dylib /opt/local/lib/libpangoft2-1.0.dylib/opt/local/lib/libpango- 1.0.dylib /opt/local/lib/libcairo.dylib -lSM -lICE /opt/local/lib/libfontconfig.dylib /opt/local/lib/libfreetype.dylib /opt/local/lib/libexpat.dylib /opt/local/lib/libpng12.dylib /opt/local/lib/libXrender.dylib -lX11 /opt/local/lib/libgnomevfs-2.dylib-lssl -lcrypto /opt/local/lib/libhowl.dylib /opt/local/lib/libgconf- 2.dylib /opt/local/lib/libORBit-2.dylib -lresolv /opt/local/lib/libgmodule- 2.0.dylib /opt/local/lib/libgpgme.dylib /opt/local/lib/libgpg-error.dylib-lldap /opt/local/lib/libsoup- 2.2.dylib /opt/local/lib/libgthread-2.0.dylib /opt/local/lib/libxml2.dylib -lpthread -lz -lm /opt/local/lib/libgnome-keyring.dylib/opt/local/lib/libavahi- client.dylib -ldl /opt/local/lib/libavahi-glib.dylib/opt/local/lib/libavahi- common.dylib /opt/local/lib/libdbus-glib-1.dylib /opt/local/lib/libdbus- 1.dylib /opt/local/lib/libgobject-2.0.dylib /opt/local/lib/libglib-2.0.dylib/opt/local/lib/libintl.dylib -lc /opt/local/lib/libiconv.dylib /usr/bin/ld: warning prebinding disabled because dependent library: /opt/local/lib/libcryptui.0.dylib is not prebound /usr/bin/ld: multiple definitions of symbol _g_free /opt/local/lib/libglib-2.0.dylib(gmem.o) definition of _g_free ../../libseahorse/libseahorse.a(seahorse-secure-memory.o) definition of _g_free in section (__TEXT,__text) /usr/bin/ld: multiple definitions of symbol _g_malloc /opt/local/lib/libglib-2.0.dylib(gmem.o) definition of _g_malloc ../../libseahorse/libseahorse.a(seahorse-secure-memory.o) definition of _g_malloc in section (__TEXT,__text) /usr/bin/ld: multiple definitions of symbol _g_malloc0 /opt/local/lib/libglib-2.0.dylib(gmem.o) definition of _g_malloc0 ../../libseahorse/libseahorse.a(seahorse-secure-memory.o) definition of _g_malloc0 in section (__TEXT,__text) /usr/bin/ld: multiple definitions of symbol _g_realloc /opt/local/lib/libglib-2.0.dylib(gmem.o) definition of _g_realloc ../../libseahorse/libseahorse.a(seahorse-secure-memory.o) definition of _g_realloc in section (__TEXT,__text) collect2: ld returned 1 exit status make[3]: *** [seahorse-preferences] Error 1 make[2]: *** [all-recursive] Error 1 make[1]: *** [all-recursive] Error 1 make: *** [all] Error 2 Warning: the following items did not execute (for seahorse): org.macports.activate org.macports.build org.macports.destroot org.macports.install Error: Status 1 encountered during processing. bash$ Does anyone have any idea to fix this? Thanks, Bob

2 3

sorry for the big source_changes email
by Anders F Björklund 15 Sep '07

15 Sep '07

Seems like the trac browser had problems with serving the proper encoding on "html.gz" files, so I thought I'd upload the web page directly... http://trac.macports.org/projects/macports/browser/users/afb/lint.html? format=raw Didn't expect "source_changes" to email the entire thing out, though. So in retrospect I should probably have kept the lint.html.gz To you reading them on dialup: my apologies :-| --anders

1 0

gcc43 installed corrupted receipt; made MacPorts unusable
by Ryan Schmidt 14 Sep '07

14 Sep '07

So we had a problem with the gcc43 port (until it was fixed in r29009) where just attempting to install gcc43 rendered the entire port system unusable, because MacPorts couldn't process the receipt that got installed. See this ticket: http://trac.macports.org/projects/macports/ticket/12698 The problem occurred when variant descriptions were added to the port like this (in r28362): variant odcctools description "some description" { [snip] } Those of us less familiar with tcl syntax rules (me; probably most of us) probably didn't catch that this should have been written: variant odcctools \ description "some description" { [snip] } or: variant odcctools description "some description" { [snip] } I don't know enough about tcl to understand why the original version was unacceptable, and in particular, if it was unacceptable, why it didn't blow up much earlier. What surprised me was that "port lint" didn't find anything wrong, and that "port install" didn't complain at all about the port, until the entire thing exploded during the install phase. "port info" gave a slight clue when it printed "(Variants: universal, darwin_8, powerpc, )" (the two variants didn't show up, but there was a comma and a blank after the powerpc variant). "port variants" similarly didn't show the two variants, but had an extra blank line at the end. What can we do to prevent portfiles malformed in this manner from breaking MacPorts in the future? It seems that there are parts of the code that are assuming things that are not true in portfiles malformed in this way. Perhaps to be more robust we should add more sanity-checking code to base. But I'm not sure where or how.

3 3