I checked out the development copy of ICU from CVS and am now able to get parrot to statically link to ICU. It however coredumps everytime parrot is invoked.
My guess is either parrot's internals are too focused on the 2.6.1 of ICU or I haven't really had as much success as I thought with ICU. Incidently, ICU still will not build cleanly on Cygwin if the --disable-shared is provided. </the sound of me punting> George R wrote: > I think the Cygwin port for using static libraries has to be finished > for ICU. The current ICU CVS snapshot should be better suited for > static libraries, but it's not completed yet. Help is always > appreciated on this matter. > You could try to set CPPFLAGS=-DU_STATIC_IMPLEMENTATION before you run > ICU. That may help to prevent the variable importation. > Of course, using asking on icu4c-support mailing list may get you a > quicker and a wider variety of responses. > PerlDiscuss - Perl Newsgroups and mailing lists wrote: > >Ok, after much gnashing of teeth, I have made *some* progress: > >1. Start as clean as possible > > A. cd /parrot > > B. make realclean > > C. cvs update -dP > > > >2. Begin rain dance > > A. modify config/gen/icu.pl (remove --disable-dynamic) > > $icuconfigureargs = "--disable-layout --disable-tests > >--disable-samples --quiet '--prefix=$cwd/blib' --enable-static > >--disable-extras '--oldincludedir=$cwd/blib/old' > >--with-data-packaging=archive"; > > B. modify config/init/hints/cygwin.pl > > link => 'c++', > > C. perl Configure.pl --ccflags=-Iicu/source/common > > Fixes a problem with #includes for unicode/<header.h> files > > D. make > > > >now fails during the linking: > > > >c++ -o parrot.exe -L/usr/local/lib -g imcc/main.o blib/lib/libparrot.a > >blib/lib/libicuuc.a blib/lib/libicuda > >ta.a -lcrypt > >blib/lib/libicuuc.a(udata.o)(.text+0x745): In function `openCommonData': > >/perl/.cpan/build/icu/source/common/udata.c:581: undefined reference to > >`__imp__icudt26_dat' > >collect2: ld returned 1 exit status > >make: *** [parrot.exe] Error 1 > > > > > >I have no idea about that error or why ICU appears to be compiled multiple > >times. > > > >Any further insight would be greatly appreciated. > > > > > >Perldiscuss - Perl Newsgroups And Mailing Lists wrote: > > > > > > > >>I am not sure if this message will make it to the list as I am posting it > >>using the www.perldiscuss.com interface. Cygwin has been failing to build > >>since at least the 21st. I suspect it has been since ICU went in. I > >>checked the tinderbox report but it is not listed. > >> > >> > > > > > > > >>FWIW - PLATFORMS is also incorrect in regards to Cygwin, using > >>--define=inet_aton is no longer required due to some hackery > >> > >> > > > > > > > >>Since I can barely spell C, I figured it would probably be best to provide > >>the entire output of make instead of just saying "it doesn't work". You > >>can find the results at > >> > >> > > > > > > > >>http://perlmonk.org/~limbicregion/cygwin.html > >> > >> > > > > > > > >>It was generated as on 2004-04-26 at 08:45AM EST. I am willing to do/try > >>anything, but anyone wishing to work with me will have to keep in mind I > >>am as green as they come ;-) > >> > >> > > > > > > > >>Cheers - Joshua > >> > >>