Re: [Fink-devel] dlopen symbol not found
On 03 Dec 2007, at 07:02, Peter O'Gorman wrote: > Jean-François Mertens wrote: >>> >> On 03 Dec 2007, at 05:35, Peter O'Gorman wrote: >> >>> >>> Did you check with otool -L that libeplplot loads libX11? >> >> But I see that Peter will give you a quicker fix - at least for >> this specific problem - the principle is the same, and if >> you want to be thorough you go through the check I >> suggested _ it doesn't take that long .. :) > > Hehe, I thought I was doing a reasonable job of giving him the > tools so > that he could learn to figure stuff out by himself :-) hoping I helped a bit too _ just thought that when the same type of problem was re-occurring with the same pkg, it was better to give some more systematic tools.. JF > > Sorry, I did not see your email to Koen about this (was it off-list?). Right _ it was _though I tried to use "Reply-to-all" throughout.. Here is what I can reconstruct (and I see it is in my first reply that I forgot the "reply-all": > > On 03 Dec 2007, at 02:52, Koen van der Drift wrote: > > >> t/1Can't load '/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- >> Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle' for module >> Bio::Emboss: dlopen(/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/ >> Bio- >> Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle, 2): Symbol >> not >> found: _XAllocColor >>Referenced from: /sw/lib/EMBOSS/libeplplot.3.dylib >>Expected in: dynamic lookup >> at t/1.t line 11 >> Compilation failed in require at t/1.t line 11. >> >> As I understand, _XAllocColor is part of libX1, and I have X11 and >> the >> X11SDK installed from the Leopard DVD. I re-installed both, rebuild >> emboss, and then tried bio-emboss-pm588 again, but still the same >> error. >> > > Try to build your pkg (adding libs if necessary to the link > commands) such > that all undefined symbols of all Mach-O files have a " (from ..." > in the output > of "nm -m" > > Jean-Francois > On Dec 2, 2007, at 9:17 PM, Jean-François Mertens wrote: > Try to build your pkg (adding libs if necessary to the link > commands) such > that all undefined symbols of all Mach-O files have a " (from ..." > in the output > of "nm -m" > Could you explain how to do that, I have no idea what you mean :-( thanks, - Koen. On 03 Dec 2007, at 03:30, Koen van der Drift wrote: > > On Dec 2, 2007, at 9:17 PM, Jean-François Mertens wrote: > > >> Try to build your pkg (adding libs if necessary to the link >> commands) such >> that all undefined symbols of all Mach-O files have a " (from ..." >> in the output >> of "nm -m" >> > > Could you explain how to do that, I have no idea what you mean :-( > dpkg -S emboss | xargs file | fgrep 'Mach-O' | cut -f1 -d: will give you the list of all binaries in your pkg Then, for each of those in turn, say $f, run nm -mgu $f | fgrep -v ' (from ' and you'll see all symbols for which it may be better to tell the linker where to find them... (to understand what this is doing, run first on some example just nm -mgu $f to see what you eliminate, and why the others are the culprits for your problems) Best, Jean-Francois __ On Dec 2, 2007, at 11:49 PM, Jean-François Mertens wrote: > But I see that Peter will give you a quicker fix - at least for > this specific problem - the principle is the same, and if > you want to be thorough you go through the check I > suggested _ it doesn't take that long .. :) > Actually there are dozes of binaries in emboss (it's a suite of programs). Only three of them have the word 'emboss' in them. If I change the test to: dpkg -L emboss | grep bin | xargs file | fgrep 'Mach-O' | cut -f1 -d: I still get a long list, assuming that I didn't make an error. So it would take some more time to go through all of them one by one :) thanks for all the help, - Koen. ___ On 03 Dec 2007, at 05:37, Koen van der Drift wrote: > > On Dec 2, 2007, at 10:05 PM, Jean-François Mertens wrote: > > >> Then, for each of those in turn, say $f, run >> >> nm -mgu $f | fgrep -v ' (from ' >> > > This line doesn't return anything. Is it complete, it looks like > there is something missing at the end? > It shouldn't return anything ! Did you try it for all binaries in the pkg ? JF _ On 03 Dec 2007, at 05:59, Koen van der Drift wrote: > > On Dec 2, 2007, at 11:49 PM, Jean-François Mertens wrote: > > >> But I see that Peter will give you a quicker fix - at least for >> this specific problem - the principle is the same, and if >> you want to be thorough you go through the check I >> suggested _ it doesn't take that long .. :) >> > > Actually there are dozes of binaries in emboss (it's a suite of > programs). Only three of them have the word 'emboss' in them. If I > change the test to: > > dpkg -L emboss | grep bin | xargs file | fgrep 'Mach-O' | cut -f1 -d: > > I still get a long list, assuming that I didn't make an error. So > it would take some m
Re: [Fink-devel] dlopen symbol not found
On Mon, Dec 03, 2007 at 12:58:18AM -0500, Daniel Macks wrote: > Hooray for upstream bugs! > > All of the shared libraried in emboss-5.0.0-3 are deficient in their > linkages. They don't link any other libraries, yet they all use > symbols from other libraries. See attached file for output from the > "fink-dyld-link-test" thing in my experimental/ directory on cvs.sf to > see what symbols are unresolved. The makefiles all appear to know the > libs that would be needed, but intentionally avoid dynamically linking > them on darwin (and don't even pass the flags correctly on most other > platforms because they didn't read the automake or libtool manual). Attached is a patchset that fixes all these linker messes. May need to redo the Depends/BuildDepends for the emboss.info package set after adding this to the existing emboss.patch file. dan -- Daniel Macks [EMAIL PROTECTED] http://www.netspace.org/~dmacks diff -Nurd -x'*~' EMBOSS-5.0.0.orig/ajax/Makefile.in EMBOSS-5.0.0/ajax/Makefile.in --- EMBOSS-5.0.0.orig/ajax/Makefile.in 2007-07-12 06:53:32.0 -0400 +++ EMBOSS-5.0.0/ajax/Makefile.in 2007-12-03 01:28:50.0 -0500 @@ -85,7 +85,8 @@ $(libajax_la_LDFLAGS) $(LDFLAGS) -o $@ @[EMAIL PROTECTED] = -rpath $(libdir) @[EMAIL PROTECTED] = -rpath $(libdir) -libajaxg_la_LIBADD = +libajaxg_la_LIBADD = libajax.la ../plplot/libeplplot.la +libajaxg_la_DEPENDENCIES = libajax.la ../plplot/libeplplot.la am__libajaxg_la_SOURCES_DIST = ajgraph.c ajhist.c @[EMAIL PROTECTED] = $(am__objects_2) libajaxg_la_OBJECTS = $(am_libajaxg_la_OBJECTS) diff -Nurd -x'*~' EMBOSS-5.0.0.orig/nucleus/Makefile.in EMBOSS-5.0.0/nucleus/Makefile.in --- EMBOSS-5.0.0.orig/nucleus/Makefile.in 2007-07-12 06:53:36.0 -0400 +++ EMBOSS-5.0.0/nucleus/Makefile.in2007-12-03 01:25:23.0 -0500 @@ -55,7 +55,8 @@ am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(includedir)" libLTLIBRARIES_INSTALL = $(INSTALL) LTLIBRARIES = $(lib_LTLIBRARIES) -libnucleus_la_LIBADD = +libnucleus_la_LIBADD = ../ajax/libajax.la +libajaxg_la_DEPENDENCIES = ../ajax/libajax.la am__objects_1 = embaln.lo embcom.lo embcons.lo embdata.lo embdbi.lo \ embdmx.lo embdomain.lo embest.lo embexit.lo embgroup.lo \ embiep.lo embindex.lo embinit.lo embmat.lo embmisc.lo \ diff -Nurd -x'*~' EMBOSS-5.0.0.orig/plplot/Makefile.in EMBOSS-5.0.0/plplot/Makefile.in --- EMBOSS-5.0.0.orig/plplot/Makefile.in2007-07-12 06:53:36.0 -0400 +++ EMBOSS-5.0.0/plplot/Makefile.in 2007-12-03 01:15:17.0 -0500 @@ -54,7 +54,7 @@ am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(pkgincludedir)" libLTLIBRARIES_INSTALL = $(INSTALL) LTLIBRARIES = $(lib_LTLIBRARIES) -libeplplot_la_LIBADD = +libeplplot_la_LIBADD = $(X_LIBS) -lX11 -lgd am__objects_1 = pdfutils.lo plargs.lo plbox.lo plcont.lo plcore.lo \ plctrl.lo plcvt.lo pldtik.lo plfill.lo plhist.lo plline.lo \ plmap.lo plot3d.lo plpage.lo plsdef.lo plshade.lo plsym.lo \ - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] dlopen symbol not found
Hooray for upstream bugs! All of the shared libraried in emboss-5.0.0-3 are deficient in their linkages. They don't link any other libraries, yet they all use symbols from other libraries. See attached file for output from the "fink-dyld-link-test" thing in my experimental/ directory on cvs.sf to see what symbols are unresolved. The makefiles all appear to know the libs that would be needed, but intentionally avoid dynamically linking them on darwin (and don't even pass the flags correctly on most other platforms because they didn't read the automake or libtool manual). dan -- Daniel Macks [EMAIL PROTECTED] http://www.netspace.org/~dmacks Preparing... g++ -all_load -multiply_defined suppress -bind_at_load /tmp/vSdsgacKlN.c -o /tmp/z82JuI69tD.o Testing package emboss-ajax5-dev... Examining /sw/lib/libajaxg.dylib... g++ -all_load -multiply_defined suppress -bind_at_load /tmp/vSdsgacKlN.c -o /tmp/z82JuI69tD.o /sw/lib/libajaxg.a /sw/lib/EMBOSS/libajaxg.5.dylib /usr/lib/libSystem.B.dylib ld: Undefined symbols: _ajAcdGetProgram _ajAcdInit _ajAcdInitP _ajCallRegister _ajCharMatchC _ajCharMatchCaseC _ajCharPrefixCaseS _ajDebug _ajDegToRad _ajErr _ajFileClose _ajFileNewOut _ajFmtPrint _ajFmtPrintAppS _ajFmtPrintF _ajFmtPrintS _ajFmtStr _ajFmtString _ajListstrFree _ajListstrGetLength _ajListstrNew _ajListstrPop _ajListstrPushAppend _ajMemAlloc _ajMemCalloc _ajMemCalloc0 _ajMemFree _ajNamInit _ajSeqGetLen _ajSeqGetNameS _ajStrAppendC _ajStrAssignC _ajStrAssignEmptyC _ajStrAssignEmptyS _ajStrAssignS _ajStrDel _ajStrGetCharLast _ajStrGetLen _ajStrGetPtr _ajStrInsertC _ajStrInsertS _ajStrKeepRange _ajStrMatchCaseC _ajStrNewC _ajTimeDel _ajTimeNewToday _ajTimeNewTodayFmt _ajUser _c_pladv _c_plbox _c_plcol0 _c_plend _c_plerrx _c_plerry _c_plfill _c_plgchr _c_plgpage _c_plinit _c_pljoin _c_plline _c_pllsty _c_plmtex _c_plpoin _c_plpsty _c_plptex _c_plschr _c_plscol0 _c_plsdev _c_plsori _c_plspage _c_plssub _c_plvpor _c_plvsta _c_plwid _c_plwind _c_plxsfnam _plsc _plstrl Examining /sw/lib/libajax.dylib... g++ -all_load -multiply_defined suppress -bind_at_load /tmp/vSdsgacKlN.c -o /tmp/z82JuI69tD.o /sw/lib/libajax.a /sw/lib/EMBOSS/libajax.5.dylib /usr/lib/libSystem.B.dylib Testing package emboss-eplplot3-dev... Examining /sw/lib/libeplplot.dylib... g++ -all_load -multiply_defined suppress -bind_at_load /tmp/vSdsgacKlN.c -o /tmp/z82JuI69tD.o /sw/lib/libeplplot.a /sw/lib/EMBOSS/libeplplot.3.dylib /usr/lib/libSystem.B.dylib ld: Undefined symbols: _XAllocColor _XAllocColorCells _XAllocNamedColor _XCheckWindowEvent _XClearWindow _XCloseDisplay _XCopyArea _XCreateColormap _XCreateFontCursor _XCreateGC _XCreatePixmap _XCreateWindow _XDefineCursor _XDestroyWindow _XDrawLine _XDrawLines _XFillPolygon _XFillRectangle _XFlush _XFree _XFreeColors _XFreeGC _XFreePixmap _XGetErrorText _XGetGeometry _XGetImage _XGetVisualInfo _XLookupString _XMapRaised _XOpenDisplay _XPending _XPutImage _XQueryColors _XQueryPointer _XSelectInput _XSetBackground _XSetErrorHandler _XSetForeground _XSetFunction _XSetLineAttributes _XSetStandardProperties _XSetWindowBackground _XSetWindowColormap _XStoreColor _XSync _XSynchronize _XUndefineCursor _XWarpPointer _XWindowEvent _gdFree _gdImageColorAllocate _gdImageColorDeallocate _gdImageCreate _gdImageCreateTrueColor _gdImageDestroy _gdImageFilledPolygon _gdImageFilledRectangle _gdImageGifPtr _gdImageLine _gdImagePngPtrEx _gdImageSetAntiAliased _gdImageSetThickness Testing package emboss-nucleus5-dev... Examining /sw/lib/libnucleus.dylib... g++ -all_load -multiply_defined suppress -bind_at_load /tmp/vSdsgacKlN.c -o /tmp/z82JuI69tD.o /sw/lib/libnucleus.a /sw/lib/EMBOSS/libnucleus.5.dylib /usr/lib/libSystem.B.dylib ld: Undefined symbols: _ajAZToInt _ajAlignDefineSS _ajAlignSetGapR _ajAlignSetMatrixFloat _ajAlignSetRange _ajAlignSetScoreR _ajDebug _ajFmtPrintAppS _ajFmtPrintF _ajMemAlloc _ajMemFree _ajMessCrashFL _ajMessSetErr _ajSeqAssignAccS _ajSeqAssignDescS _ajSeqAssignNameS _ajSeqAssignUsaS _ajSeqDel _ajSeqGapStandard _ajSeqGetAccS _ajSeqGetDescS _ajSeqGetLen _ajSeqGetNameS _ajSeqGetOffend _ajSeqGetOffset _ajSeqGetSeqC _ajSeqGetUsaS _ajSeqIsReversed _ajSeqNewNameS _ajSeqNewRangeC _ajSeqcvtGetCodeK _ajSeqstrCountGaps _ajStrAppendC _ajStrAppendK _ajStrAppendS _ajStrAssignC _ajStrAssignK _ajStrAssignS _ajStrAssignSubC _ajStrCalcCountK _ajStrDel _ajStrDelStatic _ajStrFmtUpper _ajStrGetLen _ajStrGetPtr _ajStrInsertK _ajStrNewC _ajStrReverse _ajStrSetRes _ajMemCalloc _ajMemCalloc0 _ajStrCmpS _ajStrVcm
Re: [Fink-devel] dlopen symbol not found
On 03 Dec 2007, at 05:59, Koen van der Drift wrote: > > On Dec 2, 2007, at 11:49 PM, Jean-François Mertens wrote: > >> But I see that Peter will give you a quicker fix - at least for >> this specific problem - the principle is the same, and if >> you want to be thorough you go through the check I >> suggested _ it doesn't take that long .. :) > > Actually there are dozes of binaries in emboss (it's a suite of > programs). Only three of them have the word 'emboss' in them. If I > change the test to: > > dpkg -L emboss | grep bin | xargs file | fgrep 'Mach-O' | cut -f1 -d: > > I still get a long list, assuming that I didn't make an error. So > it would take some more time to go through all of them one by one :) No .. That's why the file in question was labeled "$f" in : > dpkg -S emboss | xargs file | fgrep 'Mach-O' | cut -f1 -d: > will give you the list of all binaries in your pkg > > Then, for each of those in turn, say $f, run > > nm -mgu $f | fgrep -v ' (from ' to do in fact : for f in `dpkg -S emboss | xargs file | fgrep 'Mach-O' | cut -f1 -d:` \ do echo $f; nm -mgu $f | fgrep -v ' (from ' done Cheers, JF - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] dlopen symbol not found
> On 03 Dec 2007, at 05:35, Peter O'Gorman wrote: > > Did you check with otool -L that libeplplot loads libX11? But I see that Peter will give you a quicker fix - at least for this specific problem - the principle is the same, and if you want to be thorough you go through the check I suggested _ it doesn't take that long .. :) JF - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] dlopen symbol not found
On Dec 2, 2007, at 11:35 PM, Peter O'Gorman wrote: >> >> >> However, now I get the following, again similar error: >> >> t/1Can't load '/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- >> Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle' for module >> Bio::Emboss: dlopen(/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/ >> Bio- >> Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle, 2): Symbol >> not >> found: _XAllocColor >> Referenced from: /sw/lib/EMBOSS/libeplplot.3.dylib >> Expected in: dynamic lookup >> at t/1.t line 11 >> Compilation failed in require at t/1.t line 11. > > Did you check with otool -L that libeplplot Any suggestions on how I > can make it load libX11loads libX11? Hmm, no it doesn't: $ oTool -L libeplplot.* libeplplot.3.2.7.dylib: /sw/lib/EMBOSS/libeplplot.3.dylib (compatibility version 6.0.0, current version 6.7.0) /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 111.0.0) libeplplot.3.dylib: /sw/lib/EMBOSS/libeplplot.3.dylib (compatibility version 6.0.0, current version 6.7.0) /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 111.0.0) libeplplot.dylib: /sw/lib/EMBOSS/libeplplot.3.dylib (compatibility version 6.0.0, current version 6.7.0) /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 111.0.0) Any suggestions on how I can make it load libX11? thanks, - Koen. - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] cvs up : "Cannot allocate memory"
On 03 Dec 2007, at 05:03, Daniel Macks wrote: > Yup. Often (but sporadically) over past week or two. Certainly a > problem on SF's server end, not ours. Please file a bug with > SourceForge. Done. http://sourceforge.net/tracker/index.php? func=detail&aid=1843109&group_id=1&atid=21 JF - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] dlopen symbol not found
Koen van der Drift wrote: > On Dec 2, 2007, at 10:55 AM, Koen van der Drift wrote: >> But, I now get another similar error: >> >> t/1Can't load '/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- >> Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle' for module >> Bio::Emboss: dlopen(/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/ >> Bio-Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle, 1): >> Symbol not found: _XS_Bio__Emboss_ajStrTokenCount >> Referenced from: /sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- >> Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle >> Expected in: dynamic lookup >> at t/1.t line 11 > > > > OK, I solved that by commenting out the following line in the test file: > > BEGIN { delete $ENV{PERL_DL_NONLAZY}; }; Yes, this makes it pass RTLD_NOW to dlopen() which is similar to setting DYLD_BIND_AT_LAUNCH. It should work with this set, if it does not it means there is a missing lib or symbol when linking the loadable bundle. > > > However, now I get the following, again similar error: > > t/1Can't load '/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- > Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle' for module > Bio::Emboss: dlopen(/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- > Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle, 2): Symbol not > found: _XAllocColor >Referenced from: /sw/lib/EMBOSS/libeplplot.3.dylib >Expected in: dynamic lookup > at t/1.t line 11 > Compilation failed in require at t/1.t line 11. Did you check with otool -L that libeplplot loads libX11? Peter -- Peter O'Gorman http://pogma.com - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] cvs up : "Cannot allocate memory"
On Mon, Dec 03, 2007 at 03:55:12AM +0100, Jean-Fran?ois Mertens wrote: > I get since some time regularly failures of fink selfupdate > with msgs like > cvs update: cannot open directory /cvsroot/fink/dists/10.4/unstable/ > crypto/finkinfo: Cannot allocate memory > > I get those as well when using explicitly /usr/bin/cvs, so it is not > due to fink's cvs pkg. > This happens even when local memory usage is low _ say 1.4 out of 2 G. > > In general when retrying, especially dir after dir and with a naked > 'cvs up' without adornations, > things go away and one gets through. Tonight not... > > The ref to "/cvsroot/fink/dists/" suggests it may be either a memory > problem or an error > on the other side.. > (In general I get this on rather big dirs _ eg, 'libs' _ but tonight > even on smaller ones like crypto) > > Anyone else experiencing this ? > Is it really a sourceforge problem ? Yup. Often (but sporadically) over past week or two. Certainly a problem on SF's server end, not ours. Please file a bug with SourceForge. dan -- Daniel Macks [EMAIL PROTECTED] http://www.netspace.org/~dmacks - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] cvs up : "Cannot allocate memory"
On Dec 2, 2007, at 10:14 PM, Jean-François Mertens wrote: > > On 03 Dec 2007, at 04:03, Alexander Hansen wrote: > >> -BEGIN PGP SIGNED MESSAGE- >> Hash: SHA1 >> >> Jean-François Mertens wrote: >>> I get since some time regularly failures of fink selfupdate with >>> msgs like cvs update: cannot open directory >>> /cvsroot/fink/dists/10.4/unstable/ crypto/finkinfo: Cannot allocate >>> memory >>> >>> I get those as well when using explicitly /usr/bin/cvs, so it is >>> not due to fink's cvs pkg. This happens even when local memory >>> usage is low _ say 1.4 out of 2 G. >>> >>> In general when retrying, especially dir after dir and with a naked >>> 'cvs up' without adornations, things go away and one gets through. >>> Tonight not... >>> >>> The ref to "/cvsroot/fink/dists/" suggests it may be either a >>> memory problem or an error on the other side.. (In general I get >>> this on rather big dirs _ eg, 'libs' _ but tonight even on smaller >>> ones like crypto) >>> >>> Anyone else experiencing this ? Is it really a sourceforge problem >>> ? >>> >>> JF >>> >>> PS: crypto finally got through, after ~8 iterations >> I have, and others have talked about it on IRC. > > And ? _ the conclusion of the talking ? > > JF Unfortunately the usual conclusion: sourceforge has issues that aren't communicated well or resolved quickly. Dave -- David Reiser [EMAIL PROTECTED] - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] cvs up : "Cannot allocate memory"
On 03 Dec 2007, at 04:03, Alexander Hansen wrote: > -BEGIN PGP SIGNED MESSAGE- > Hash: SHA1 > > Jean-François Mertens wrote: >> I get since some time regularly failures of fink selfupdate with >> msgs like cvs update: cannot open directory >> /cvsroot/fink/dists/10.4/unstable/ crypto/finkinfo: Cannot allocate >> memory >> >> I get those as well when using explicitly /usr/bin/cvs, so it is >> not due to fink's cvs pkg. This happens even when local memory >> usage is low _ say 1.4 out of 2 G. >> >> In general when retrying, especially dir after dir and with a naked >> 'cvs up' without adornations, things go away and one gets through. >> Tonight not... >> >> The ref to "/cvsroot/fink/dists/" suggests it may be either a >> memory problem or an error on the other side.. (In general I get >> this on rather big dirs _ eg, 'libs' _ but tonight even on smaller >> ones like crypto) >> >> Anyone else experiencing this ? Is it really a sourceforge problem >> ? >> >> JF >> >> PS: crypto finally got through, after ~8 iterations > I have, and others have talked about it on IRC. And ? _ the conclusion of the talking ? JF - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] cvs up : "Cannot allocate memory"
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Jean-François Mertens wrote: > I get since some time regularly failures of fink selfupdate with > msgs like cvs update: cannot open directory > /cvsroot/fink/dists/10.4/unstable/ crypto/finkinfo: Cannot allocate > memory > > I get those as well when using explicitly /usr/bin/cvs, so it is > not due to fink's cvs pkg. This happens even when local memory > usage is low _ say 1.4 out of 2 G. > > In general when retrying, especially dir after dir and with a naked > 'cvs up' without adornations, things go away and one gets through. > Tonight not... > > The ref to "/cvsroot/fink/dists/" suggests it may be either a > memory problem or an error on the other side.. (In general I get > this on rather big dirs _ eg, 'libs' _ but tonight even on smaller > ones like crypto) > > Anyone else experiencing this ? Is it really a sourceforge problem > ? > > JF > > PS: crypto finally got through, after ~8 iterations I have, and others have talked about it on IRC. -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFHU3H3B8UpO3rKjQ8RArXqAJ9LEO5iTyPe6llFsDQLGEJJywGoSQCfbSB8 ojF+TIcycqU4afzigWfiH8s= =yNIT -END PGP SIGNATURE- - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
[Fink-devel] cvs up : "Cannot allocate memory"
I get since some time regularly failures of fink selfupdate with msgs like cvs update: cannot open directory /cvsroot/fink/dists/10.4/unstable/ crypto/finkinfo: Cannot allocate memory I get those as well when using explicitly /usr/bin/cvs, so it is not due to fink's cvs pkg. This happens even when local memory usage is low _ say 1.4 out of 2 G. In general when retrying, especially dir after dir and with a naked 'cvs up' without adornations, things go away and one gets through. Tonight not... The ref to "/cvsroot/fink/dists/" suggests it may be either a memory problem or an error on the other side.. (In general I get this on rather big dirs _ eg, 'libs' _ but tonight even on smaller ones like crypto) Anyone else experiencing this ? Is it really a sourceforge problem ? JF PS: crypto finally got through, after ~8 iterations - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] dlopen symbol not found
On Dec 2, 2007, at 10:55 AM, Koen van der Drift wrote: > > On Dec 2, 2007, at 1:07 AM, Peter O'Gorman wrote: > >> >> I suggest that you find out what plsc is (rebuild the package that >> makes >> libajaxg.5.dylib, and grep for it) and why it is not available (what >> provides it? grep -rl plsc /sw/lib > > Thanks Peter, > > The grep command revealed two broken aliases. One of them was from > the emboss package, which I fixed, and I don't get the error > anymore. The other one was lebecpg.4.dylib, which seems to belong to > postgresql. I think I once installed it, but right now I don't have > that installed anymore. Not sure why it is still on my system and > why it would contain plsc, maybe just a coincidence. > > But, I now get another similar error: > > t/1Can't load '/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- > Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle' for module > Bio::Emboss: dlopen(/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/ > Bio-Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle, 1): > Symbol not found: _XS_Bio__Emboss_ajStrTokenCount > Referenced from: /sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- > Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle > Expected in: dynamic lookup > at t/1.t line 11 OK, I solved that by commenting out the following line in the test file: BEGIN { delete $ENV{PERL_DL_NONLAZY}; }; This does the following (see http://www.perl.com/lpt/a/520): [quote]... it passes a flag to dlopen which attempts to ensure that all functions are relocated as soon as the shared object is loaded. Sounds complicated? In the normal, "lazy" operation of the dynamic loader, the loader doesn't actually load all the functions from the library file into memory at one go - instead, it merely notices that it has a bunch more functions available; when a function is called, it loads up the appropriate part of the object into memory, and jumps to it. (Not entirely unlike the behaviour of use autouse or AutoSplit.) Setting [PERL_DL_NONLAZY] forces the loader to load up all functions at once, so that it can ensure that it really does have code for all the functions it claims to have code for; this is usually what you want to do when testing.[/quote] However, now I get the following, again similar error: t/1Can't load '/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle' for module Bio::Emboss: dlopen(/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle, 2): Symbol not found: _XAllocColor Referenced from: /sw/lib/EMBOSS/libeplplot.3.dylib Expected in: dynamic lookup at t/1.t line 11 Compilation failed in require at t/1.t line 11. As I understand, _XAllocColor is part of libX1, and I have X11 and the X11SDK installed from the Leopard DVD. I re-installed both, rebuild emboss, and then tried bio-emboss-pm588 again, but still the same error. Any clues? thanks, - Koen. - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] Failed: phase compiling: oleo-1.99.16-1021 failed
On Sat, Dec 01, 2007 at 06:20:00PM +0100, Dominique Dhumieres wrote: > Installing oleo-1.99.16-1021 on Intel 10.5.1 failed with: > > ... > checking for IceConnectionNumber in -lICE... no > checking for Motif... libraries /sw/lib, headers in default path > configure: error: Xbae not found; you must install it or use --without-motif > checking for Xbae... libraries test link failed, headers test link failed [...] > Looking at config.log, I see a lot of > > error: Not a directory - unable to create 'conftest.dSYM/Contents' bundle > directory. Those are normal on Leopard. > Note that I checked that Xbae is installed. The important part of config.log is the section dealing with the actual Xbae test. The first relevant line in that file is probably: configure:3466: checking for Xbae dan -- Daniel Macks [EMAIL PROTECTED] http://www.netspace.org/~dmacks - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
[Fink-devel] Failed: phase compiling: oleo-1.99.16-1021 failed
Installing oleo-1.99.16-1021 on Intel 10.5.1 failed with: ... checking for IceConnectionNumber in -lICE... no checking for Motif... libraries /sw/lib, headers in default path configure: error: Xbae not found; you must install it or use --without-motif checking for Xbae... libraries test link failed, headers test link failed ### execution of ac_cv_lib_mysqlclient_mysql_connect=yes failed, exit code 1 Removing runtime build-lock... Removing build-lock package... /sw/bin/dpkg-lockwait -r fink-buildlock-oleo-1.99.16-1021 (Reading database ... 122679 files and directories currently installed.) Removing fink-buildlock-oleo-1.99.16-1021 ... Failed: phase compiling: oleo-1.99.16-1021 failed Looking at config.log, I see a lot of error: Not a directory - unable to create 'conftest.dSYM/Contents' bundle directory. Note that I checked that Xbae is installed. TIA Dominique - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] Failed: phase compiling: xmhtml-1.1.7-4 failed on Leopard
> I have made a new xmhtml.patch file that fixes this. It does!-) Thanks for the patch. Dominique - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] dlopen symbol not found
On Dec 2, 2007, at 1:07 AM, Peter O'Gorman wrote: > > I suggest that you find out what plsc is (rebuild the package that > makes > libajaxg.5.dylib, and grep for it) and why it is not available (what > provides it? grep -rl plsc /sw/lib Thanks Peter, The grep command revealed two broken aliases. One of them was from the emboss package, which I fixed, and I don't get the error anymore. The other one was lebecpg.4.dylib, which seems to belong to postgresql. I think I once installed it, but right now I don't have that installed anymore. Not sure why it is still on my system and why it would contain plsc, maybe just a coincidence. But, I now get another similar error: t/1Can't load '/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle' for module Bio::Emboss: dlopen(/sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle, 1): Symbol not found: _XS_Bio__Emboss_ajStrTokenCount Referenced from: /sw/src/fink.build/bio-emboss-pm588-5.0.0.1-1/Bio- Emboss-5.0.0.1/blib/arch/auto/Bio/Emboss/Emboss.bundle Expected in: dynamic lookup at t/1.t line 11 Using the grep command, ajStrTokenCount is in /sw/lib/libajax*, which is installed. I'll have another look at it later today. cheers, - Koen. - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] problem with octave ?
The problem is purely within octave; nothing to do with octave-forge. Octave shows the same seg-fault on its own test-suite, cf http://www.cae.wisc.edu/pipermail/bug-octave/2007-November/003970.html Jean-Francois On 01 Dec 2007, at 14:22, ALexandre Vial wrote: > Jean-François Mertens uclouvain.be> > writes: > >> >> Putting this to fink-devel... >> >> Since I won't have time to delve into this for the next couple of >> days, >> please have a look at tracker item >> https://sourceforge.net/tracker/? >> func=detail&atid=414256&aid=1825780&group_id=17203 >> >> JF Mertens > > > I have been playing with all of this for afew days on both Intel and > PPC, with Mac OS X 10.4.11 > > About octave-forge : there should be a build-depend on libssh2 or > libssh2.1, as it is required for building package octcdf. > > What I have tested: > > Set 1: > octave 2.9.17-1 > octave-forge 20071014-2 > (always built with g95) > > Set 2: > octave2.9.17-2 > octave-forge 20071014-4 > (built with g95 or gfortran depending on what is installed) > > Set 1 works on both Intel and PPC, and is what people should use > for now > on Mac OS X 10.4. > > For Set 2, if octave is built WITHOUT gfortran (i.e. if it is not > installed, even with ConfigureParams: --with-f77=%p/bin/gfortran, but > g95 is installed), then everything is fine, BUT building octave-forge > writes directly in /sw/share /octave, wich is not good, as the .deb > does > not contain anything actually, so it is better to stay with rev. 2 of > this package. > Now if gfortran is installed, octave compiles fine, but I get a > seg-fault when building octave-forge on Intel, I did not try on PPC, > maybe next week. > - SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel