Re: AWT on DOS platform
On May 29, 2001, "David Meggy" <[EMAIL PROTECTED]> wrote: > On this website http://www.transvirtual.com/kaffe-awt.htm is a > description of AWT runnning on a platform (DOS) with Java running at > a level just above the hardware, and a screenshot. However when I > downloaded kaffe, all I can find is win32 and X. Where is this code > that runs on DOS? AFAIK, it was never integrated in the public version of Kaffe. TVT used to sell it commercially, but I heard it was dead a while ago. Tim, please correct me if I'm mistaken. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: newbie:ClassNotFoundException when running application
On May 28, 2001, Frederic ALBERT <[EMAIL PROTECTED]> wrote: > Here is my CLASSPATH : > /tests/java/TextEdit.class:/tests/java/kaffe/test/awt/Calc/Calc.class: No point in mentioning .class files in CLASSPATH. Only directories and .jar/.zip archives. It seems that your problem is that you're missing `.' somewhere in the CLASSPATH. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: What's this error?Re: Hi,everyone here! About Kaffe on Linux-vr
On Apr 18, 2001, [EMAIL PROTECTED] wrote: > I still can not get the Kaffe cross-compiled. The tests done by configure that try to detect the size of certain data types won't work when cross compiling. You'll have to set them by hand, or rebuild configure using a recent pre-release of autoconf (2.49d, for example), in which these tests are done with a compile test (that doesn't require running a program, something almost impossible to do in a cross-compilation environment. The variables you'll want to set in config.cache or in your environment are ac_cv_sizeof_{short,int,long,long_long,__int64,voidp}. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: What's this error?Re: Hi,everyone here! About Kaffe on Linux-vr
On Apr 16, 2001, [EMAIL PROTECTED] wrote: > /usr/bin/mipsel-linux-ar cr cru .libs/libltdlc.al ltdl.lo Looks like you've messed up either AR or ARFLAGS: there should only be one of `cr' or `cru'. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: Linking problems with Kaffe VM
On Nov 24, 2000, Peter Eisentraut <[EMAIL PROTECTED]> wrote: > (b) On ELF systems ISTM that you could link Kaffe with -Bsymbolic. This > is not infinitely portable though. Yep. On systems that support it, adding -Wl,-Bsymbolic to the link command should be enough. I don't think there's any alternate spelling for this switch, so it's kind of pointless to support it in libtool. Well, it could always discard the switch on platforms in which it isn't supported, but I have trouble trying to convince myself it's any better than having to explicitly test for it, and possibly bail out when it's not supported. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: JNI Problem
On Sep 14, 2000, "Prashant Patel" <[EMAIL PROTECTED]> wrote: >gcc -c -fPIC Hello.c -o Hello.o >ld -shared -o libHello.so Hello.o The notion of shared library extension varies from system to system. Also, in some cases, shared libraries have to be linked with libgcc in order to get some C support functions. These are two potential problem spots, that you're saved from if you use libtool to create your shared libraries. However, since you don't, check that libltdl is using the correct shared library extension (look for LTDL_SHLIB_EXT in libltdl/config.h, in the build tree) and try to use GCC for linking. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: TestNative.c
On Aug 17, 2000, "Johan Andersson" <[EMAIL PROTECTED]> wrote: > What is the purpose of the functions in /libraries/clib/native/TestNative.c? > When are the functions used? test/regression/TestNative.java -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: missing files, while compile Kaffe
On Aug 14, 2000, Mansuroglu Seki <[EMAIL PROTECTED]> wrote: > - cd to 'd:\cygwin\kaffe\kaffe\kaffe' Wrong. Run make in the same directory in which you run configure. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: Classpath and Kaffe
On Jun 28, 2000, "Nic Ferrier" <[EMAIL PROTECTED]> wrote: > I'm confused about TranV's position now... the Kaffe web site says > that no code contributed to Kaffe will be used by TransV. I don't see > why not... I certainly wouldn't have a problem with it - they're > supporting the GPL and I applaud that and want to help. They must be able to license the code to their customers under different licenses. Same issue with libgcj: Cygnus (now Red Hat) took a different approach of requiring copyright assignments, just like FSF, in order to integrate changes from network contributors. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC Developer aoliva@{cygnus.com, redhat.com} CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org} Free Software Evangelist*Please* write to mailing lists, not to me
Re: Cross-compliling for Kaffe for ARM-LINUX
On May 24, 2000, Euisung Kang <[EMAIL PROTECTED]> wrote: > However, when I compile a java program (e.g. #javac hello.java), > I get the following error message. I am trying to correct this problem. > Any help? It's not just Kaffe that must be compiled with -fPIC. In fact, it's alreayd compiled with -fPIC by default. It's any other libraries that Kaffe uses, such as zlib, gmp, etc (see FAQ/FAQ.requiredlibraries) that must be shared libraries or at least built with -fPIC. The alternative is to upgrade glibc to some recent release that doesn't present this dynamic linker bug. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: HELP !!! Install kaffe on Windows NT4
On May 19, 2000, "Baudino, Gilles" <[EMAIL PROTECTED]> wrote: >checking whether the C compiler (gcc ) works... no >configure: error: installation or configuration problem: C compiler > cannot create executables. > What have I to make to run the configure script ? Make sure GCC is working, for a start. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: none
On May 15, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > However, Godmar has pointed out to me that Kaffe's current > implementation is within the specification (the spec makes no > fairness or liveness guarantees). Additionally, Kaffe breaks broken > code "faster". If you really want fairness, then you'll have to > implement a queue of some sort above the Java primitives. Agreed. I think we can keep the code to enable fairness `#ifdef'ed out for whoever needs it, but, by default, we should try to be as efficient as possible, within the bounds of the specification. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: How to utilize kaffe win32 thread design?
On Apr 19, 2000, Jim King <[EMAIL PROTECTED]> wrote: > Do anyone know the reasons and how to utilities this Win32 system? I think you'd have to build with a native (non-cygwin/mingwin) Win32 compiler. But the `configure/make/make install' build doesn't support that. IIRC, the makefiles or equivalent for building with a native compiler are available somewhere. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: FW: Kaffe 1.0.5 build failure - NetBSD 1.4.2 (mips,pmax)
On Apr 19, 2000, Godmar Back <[EMAIL PROTECTED]> wrote: >> >> Based on the attached email from Alexandre Olivia on the Kaffe mailing list, >> I applied the following changes to the top level (/pkgsrc/lang/kaffe) >> Makefile: > Always keep in mind that changes to the Makefiles don't survive a new > configure since configure rebuilds Makefiles from Makefile.ins. I believe that file is outside the build tree. Must be something like FreeBSD's Ports collection. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: Kaffe 1.0.5 build failure - NetBSD 1.4.2 (mips,pmax)
On Apr 14, 2000, "Robert D. Mohr" <[EMAIL PROTECTED]> wrote: > I gather from past posts to this list that the JIT is not currently > supported on the mips platform. I think that's correct. At least, not on all mips platforms. Run configure --with-engine=intrp and it should work. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: General question
On Apr 11, 2000, Archie Cobbs <[EMAIL PROTECTED]> wrote: > Julien Dumesnil writes: >> Is it more advanced than the GPL version? Or is it a stable, thus older, >> version of the opensourced version we all have? > No, it's a clean room implementation. >> The bottom line being "are they completely different products?". > Yes. I don't think they're that different, but then, I've never looked at the Custom Edition code. It must not be that difference, since Tim imports (part of?) the Custom code into the open version every now and then. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: beos-native port [Was: Re: changes to thread locking layer]
On Mar 18, 2000, Alexandre Oliva <[EMAIL PROTECTED]> wrote: > On Mar 17, 2000, "Alan L. Batongbacal" <[EMAIL PROTECTED]> wrote: >> Note the two "-lz" occurrences (a bug in configure.in, maybe? > Nope, that would be a bug in libtool, if there's any bug at all. There was a bug, after all. Here's a fix. I'll try to get it into the libtool CVS tree and update Kaffe's view of libtool afterwards. Please let me know if it fixes the problem for you. Index: ltmain.sh === RCS file: /cvs/kaffe/kaffe/ltmain.sh,v retrieving revision 1.33 diff -u -r1.33 ltmain.sh --- ltmain.sh 2000/02/25 02:28:35 1.33 +++ ltmain.sh 1999/12/10 19:03:28 @@ -1570,7 +1570,6 @@ if test $linkmode = prog && test $pass != link; then new_lib_search_path="$new_lib_search_path $ladir" - deplibs="$lib $deplibs" linkalldeplibs=no if test "$link_all_deplibs" != no || test "$fast_install" != no || \ @@ -1596,6 +1595,8 @@ esac tmp_libs="$tmp_libs $deplib" done + + deplibs="$lib $deplibs" continue fi -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: beos-native port [Was: Re: changes to thread locking layer]
On Mar 17, 2000, "Alan L. Batongbacal" <[EMAIL PROTECTED]> wrote: > Note the two "-lz" occurrences (a bug in configure.in, maybe? Nope, that would be a bug in libtool, if there's any bug at all. Edouard recently reported a similar problem, but starting the build from a clean tree fixed it up. Probably it was just some remnant from the buggy libtool that lived in the Kaffe CVS tree for a few days, fixed with the update that shortly followed. If you still get the problem, after starting with a clean tree, please let me know. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: patch for a verbose -fullversion
On Mar 13, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > Its somewhat annoying, but in reality the generation and recompile are > lost in the shuffle of symbol extraction and relinking. I see :-) But that's not the point. Standards say that, when you run `make; make install', `make install' mustn't do anything. That's because many people configure/build as one user and `make install' as another. In this case, relinking the executable would be exactly the wrong thing to do. And then, there are `make's that don't support .PHONY, so you can't get what you want that way. I'm removing the .PHONY part and replacing it with a dependency on ChangeLog and config.status, as I had suggested, so that we comply with standards or, at least, with users' expectations. If you want this file to be updated every time you compile, you can hack your local copy of Kaffe to do it. I believe most people will prefer that executables aren't rebuilt every time they `make'. > Done. I also did the '-rm' thing and used a temp file for the > output. > For those who are curious, here's a sample of 'kaffe -fullversion': Thanks, it's really great! :-) -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: state of the developer tools?
On Mar 13, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > Excellent. I snapped up your automake snapshot installed it and > re-built Kaffe's Makefile.in's and they're no different from the > checked in ones. That's because I've already checked in the files as produced by the new automake :-) > I assume this warning from automake: > test/regression/Makefile.am:19: warning: automake does not support conditional >definition of TEST_Overflow in TESTS > isn't a problem. Right? Right. The warning is bogus; the generated Makefile.in is correct. -- Alexandre OlivaEnjoy Guaraná, see http://www.ic.unicamp.br/~oliva/ Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: state of the developer tools?
On Feb 27, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: >> Dunno, I haven't tested it lately. If you give it a try and it >> works for you, just `make dist' it (automake CVS) and let us know. > I tried it out and it works (as well as I can tell). I looked over > the Makefile.in diffs, and there are some bug fixes, but it doesn't > look like there should be any problems. Indeed, thanks. I wonder which version of autoconf you've used, though. Since 1999-12-09, automake wouldn't have worked with autoconf 2.13, the latest stable release, because of some a quoting problem related with backticks in error messages (hi, Pavel :-). I've just fixed automake to work with both autoconf 2.13 and CVS autoconf, created a new snapshot, updated the FAQ and committed the new Makefiles. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: patch kaffe main.c for 2000
On Mar 10, 2000, Archie Cobbs <[EMAIL PROTECTED]> wrote: > Alexandre Oliva writes: > I'm no lawyer, but that seems unnecessary.. everything I've ever > seen uses ranges.. there's no ambiguity (unless you're a lawyer I guess). Yep, that's the point. But then, it's up to TVT to decide how to protect their copyright :-) -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: patch for a verbose -fullversion
On Mar 10, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > -Kaffe_SOURCES = main.c > +Kaffe_SOURCES = main.c version.o ^ why not version.c ? Since version-info.h is created, you should probably add a dependency of version.o on it. Or you could simply not create version.c, and just add the #include and the new functions to main.c. It's small enough to afford re-compilation every time. I don't like having phony targets that cause something to be built every time I run `make'. How about arranging for version-info.h to depend on config.status and $(srcdir)/ChangeLog ? You could extract the first line of ChangeLog like this: version-info.h: config.status $(srcdir)/ChangeLog [...] sed < $(srcdir)/ChangeLog \ -e 's/^/#define VER_CHANGELOG_HEAD "/' \ -e 's/$/"/' -e '1q' >> $@T [...] -rm -f $@ mv $@T $@ Note the use of a temporary file to hold the output, so that you never use a partially-constructed file. And note the use of `-' before the `rm'; on some platforms, even `rm -f' may fails. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: patch kaffe main.c for 2000
On Mar 9, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > - fprintf(stderr, "Copyright (c) 1996-1999\nTransvirtual >Technologies, Inc. All rights reserved\n"); > + fprintf(stderr, "Copyright (c) 1996-2000\nTransvirtual >Technologies, Inc. All rights reserved\n"); FYI, the FSF lawyer recommends listing all the years in which changes have been made, not just an interval, that is ambiguous. So, he'd suggest writing 1996,97,98,99,2000 (yes, omitting the century for intermediate years is fine). -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: sysdepCallMethod for RISC - request for comments
On Mar 2, 2000, Pavel Roskin <[EMAIL PROTECTED]> wrote: > return longSysdepCallMethod(call, args, func, callargs, calltype) > and gcc warns: > `return' with a value, in function returning void > Is it an error in your implementation? Probably a thinko, unless the return types changed while I wasn't looking :-) Feel free to replace: return longSysdepCallMethod(...); with longSysdepCallMethod(...); return; > Also the comment says that longSysdepCallMethod cannot be inlined, but it > is declared "inline" The comment also says (at least on alpha) that some day GCC may be able to inline these, and then, we'll want them to be inlined. > By the way, what's the reason to use "inline"? And what's the reason to > use #define for other architectures? Is it only about speed? Yep, it's only about speed. > Maybe somebody could clean up sysdepCallMethod for Alpha so that I could > use it as a base for a better implementation for PowerPC (and eventually > PA-RISC)? What do you mean with `clean it up'? Address the issues you've raised here or something else? -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: state of the developer tools?
On Feb 27, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > Is the CVS version of automake still missing features that Kaffe > requires, or could we all synchronize on that? Dunno, I haven't tested it lately. If you give it a try and it works for you, just `make dist' it (automake CVS) and let us know. It's probably working for the uses Kaffe makes of it, but some new features are likely to be unstable. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: Kaffe requires -g
On Feb 24, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > -O1 fails Which platform? Which compiler? What is the symptom? -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: Problems with awt over Kaffe 1.0.4
On Feb 22, 2000, Rodrigo Dias Arruda Senra <[EMAIL PROTECTED]> wrote: > In my case the problem was that libawt* was not built at all! The reason is pretty obvious to me: configure:3416: checking for X configure:3483: gcc -E -I/usr/local/include conftest.c >/dev/null 2>conftest.out configure:3479: X11/Intrinsic.h: No such file or directory configure: failed program was: #line 3478 "configure" #include "confdefs.h" #include configure:3559: gcc -o conftest -g -O2 -Wall -Wstrict-prototypes -I/usr/local/include conftest.c -lXt -L/usr/local/lib 1>&5 configure:3554: warning: function declaration isn't a prototype configure: In function `main': configure:3555: warning: implicit declaration of function `XtMalloc' /usr/bin/ld: cannot find -lXt collect2: ld returned 1 exit status configure: failed program was: #line 3552 "configure" #include "confdefs.h" int main() { XtMalloc() ; return 0; } Do you have X11 headers and libraries installed somewhere gcc can find them? > Peter DeLaurentis wrote: >> it went ok BUT for AWT. I get a similiar error. I cannot compile >> applications with AWT components with KJC - but this does not >> overly concern me since I have another java compiler. AFAIK, you should also install XFree86 for cygwin, otherwise Kaffe's libawt won't be built. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: Problems building Kaffe 1.0.5 on Irix 6.5
On Feb 16, 2000, "Bishop, Michael M" <[EMAIL PROTECTED]> wrote: > While compiling kaffe/kaffevm/jit3/funcs.c, I received several messages > stating that NR_ARGUMENTS and NR_REGISTERS were undeclared. Yep, I've observed that too. I'm not sure the jit3 is supposed to work on IRIX, particularly IRIX 6. Meanwhile, I've been using --with-engine=intrp. > The assembler being used is /usr/bin/as. I can't > use the GNU assembler because it doesn't accept the -n32 argument. Maybe there's some -mgnu-as flag that would allow you to tell GCC that the assembler to be used is GNU as. If there isn't, you'd have to configure and rebuild *gcc* --with-gnu-as. Another option is to write a wrapper script called `as', that removes the `-n32' switch from the command line and calls GNU as. You'd have to point gcc to it with -B/the/directory/that/contains/ the script (note trailing slash!) -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: Strange things with the current CVS
On Feb 16, 2000, Pavel Roskin <[EMAIL PROTECTED]> wrote: >> Does "current CVS" include the libtool changes checked in yesterday? > Yes, of course. That's why I'm addressing you. >> Unfortunately, I don't understand much about libtool. But Alexandre >> should be able to help out. > Actually, I should be able too :-) Thanks. Unfortunately, I'm about to leave for a 4-days trip, so I can't do much hacking right now. I'll be reading e-mail every now and then, but not that often :-( It seems that there's something very wrong with CVS libtool :-( > I just wanted to alert interested parties before I find time to > solve this problem. Please let people in the libtool mailing list know about it. Hopefully some of them may be able to help you work it out while I'm away. Thanks. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva/ Enjoy Guaraná Cygnus Solutions, a Red Hat companyaoliva@{redhat, cygnus}.com Free Software Developer and EvangelistCS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org} Write to mailing lists, not to me
Re: DOS drive names
On Feb 7, 2000, [EMAIL PROTECTED] (Marcus G. Daniels) wrote: > This isn't suitable as a patch, but here are two places where > DOS-style pathnames (C:\path\file) aren't handled, and spooky things > happen as a result... But, with your patch, it would say `C:file' is absolute. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: large Class.forName() patch
On Feb 4, 2000, Mo DeJong <[EMAIL PROTECTED]> wrote: > The current Kaffe implementation means that a \0 embedded in the > class name will screw up code that expects a \0 at the end of a > string. AFAIK, there can't be \0s embedded in class names. And, in any case, utf8 strings won't contain a \0 char unless they actually contain \u0. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: crash loading gcj stuff
On Feb 2, 2000, Godmar Back <[EMAIL PROTECTED]> wrote: > It would suck if RedHat Linux installation by virtue of including gcc > would lock you into using a particular run-time for compiled java code. > (I don't think that's the case here though) As long as you use gcc for linking, as opposed to gcj, there's no reason to expect this kind of problem. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: BUG in ZipInputStream
On Jan 27, 2000, Hiroshi Oota <[EMAIL PROTECTED]> wrote: > I built Kaffe from CVS tree today under FreeBSD-3.4-RELEASE. > The ChangeLog says >Tue Dec 28 10:14:16 CET 1999 Edouard G. Parmelan <[EMAIL PROTECTED]> Do you mean this is the latest entry in the ChangeLog? It's more than a month old, and there have been dozens of changes after that. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: File.getAbsolutePath() bugfix
On Jan 14, 2000, "Edouard G. Parmelan" <[EMAIL PROTECTED]> wrote: > Alexandre Oliva wrote: >> BTW, waitpid doesn't load pid either :-( > Okay, I will fix that too Thanks -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: File.getAbsolutePath() bugfix
On Jan 14, 2000, "Edouard G. Parmelan" <[EMAIL PROTECTED]> wrote: > I see, pid is not saved after fork. > Fixed now, thanks. BTW, waitpid doesn't load pid either :-( -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Regression Test failures in current CVS
On Jan 14, 2000, Dalibor Topic <[EMAIL PROTECTED]> wrote: > Adding a LC_ALL=C in the TestScript would be nice, it > is not set there at all. Thanks. I'm arranging that both LC_ALL and LANG are reset to C if they're previously set. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Regression Test failures in current CVS
On Jan 14, 2000, Dalibor Topic <[EMAIL PROTECTED]> wrote: > I assume that the problem lies in the upgrade from > textutils 1.22 (used in RH 6.0) and textutils 2.0 (RH > 6.1), since all failing tests use "// Sort output", > and in the case I inspected closely, Reflect.java, the > only differences were in the order of a few lines in > the sorted output. I'va attached the files > Reflect.fail and Reflect.out to this mail. > Did anyone else experience something similar? Nope, but I've also got textutils 2.0. What is your LANG environment variable set to? -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Regarding Quick opcodes
On Jan 11, 2000, Sudheer Hegde <[EMAIL PROTECTED]> wrote: > I'd like to know whether Kaffe uses the QUICK mechanism and if so It doesn't. AFAIK, this mechanism is patented. Moreover, _QUICK instructions cannot be used in a classfile, and the JIT makes them useless. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: building kaffeh when cross-compiling
On Jan 10, 2000, "btlooi" <[EMAIL PROTECTED]> wrote: >i plan using cygwin to compile the kaffe for the windows platform but > anyone can tell me where I can get a copy of pthread library.. You don't need it. Kaffe implements its own threading library. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: CVS server down?
On Jan 9, 2000, Dalibor Topic <[EMAIL PROTECTED]> wrote: > Is it just me, or is the CVS server down right now? It's probably caused by a DNS problem Transvirtual has been facing lately :-( -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: building kaffeh when cross-compiling
On Jan 7, 2000, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > Kaffeh is built when cross-compiling Kaffe. There is some magic in > configure that tries to only build kaffeh if kaffe is not being > cross-compiled. Not really. It just forces kaffeh to be built before it is used. But, as you pointed out, when cross-compiling, we can't use the kaffeh we'd build, because it won't run on the build machine, so we currently require a locally-executable kaffeh to be specified in an environment variable. > However, kaffeh is built by default, anyway! Yes, because, even though it is also needed during the build, it *is* a development tool of the kaffe suite. Just like we install kaffec, we install kaffeh, so that one doesn't need the Kaffe build-tree to be able to generate C headers for Java classes. It is possible to arrange that another version of kaffeh is built with an alternate compiler, but it requires a lot of hacking, since a separate config.h must be generated for this other compiler. The easiest way to do it would be to have a separate configure script for kaffeh. When Godmar first detected this problem, we decided this would be overkill, and agreed to require a pre-installed kaffeh for cross-builds. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: install errors
On Dec 15, 1999, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > Archie Cobbs wrote: >> Try completely blowing away your build directory hierarchy and >> starting over.. this is what I do, > Yeah, that's what I end up doing from time to time. But, I really > rather not have to do that each time I change a header file. You shouldn't have to. Please give it one more try, I've just fixed the libtool bug that was causing the weirdness at install time. The libtool snapshot I had installed in the Kaffe tree was much less stable than I had expected, sorry. It should be much better now. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: install errors
On Dec 14, 1999, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > With a very fresh snapshot, I'm getting errors during kaffe's install. How fresh is that? Archie has worked around this libtool problem last Thursday. But you may have to remove all .la files in your build tree before his change takes effect. > I think the problem might be the '(' and ')' passed to > -export-symbols-regexp. Yup, libtool doesn't quote them properly, so they have to be double-quoted. -- Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: newbie classpath problem
On Sep 8, 1999, "Frank Morton" <[EMAIL PROTECTED]> wrote: > source='inflate.c' object='inflate.o' @AMDEPBACKSLASH@ > /bin/sh: @AMDEPBACKSLASH@: command not found Make sure no Makefile.in, configure or aclocal.m4 in your source tree is different from the files in the CVS repository (cvs update should do the trick, assuming you have downloaded the snapshot with CVS). If you downloaded the tarball, you could start by touching, in this sequence, aclocal.m4, configure, config/stamp-h1.in, include/stamp-h2.in and then all the `Makefile.in's. This should prevent any configuration file from being rebuilt locally. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Problems w. Compiling on WinNT
On Aug 17, 1999, [EMAIL PROTECTED] wrote: > Just another thing: The STATUS_ACCESS_VIOLATION is caused by sh.exe. Could > it be possible that the B20.1 version of sh.exe does not work properly on > my Windows system(s)? Could be. Libtool sometimes asks too much from shells :-( What if you run `make CONFIG_SHELL="/bin/sh -x"', where does it crash? > And a last one: What about the locale settings? Maybe the make process > succeeds after switching to US locale settings? What settings do you have? None at all. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Problems w. Compiling on WinNT
On Aug 17, 1999, [EMAIL PROTECTED] wrote: >> Weird, I've just built it on Cygwin 20.1/WNT Server 4.0SP4 (maybe >> SP3?), on a 486!!!, and it appears to be working fine. Which versions >> (Cygwin, WNT) are you using? > I tried to compile the latest kaffe snapshot with Cygwin B20.1 and the > Cygwin snapshot from January on two different Windows platforms and got > always the same error message. > - Windows NT 4.0 Workstation with SP3 (Build 1381) on Pentium II/350 That's exactly the one I've got, except that it's NT Server and it's mostly in Portuguese. > - Windows 95 B on Pentium 150 > Maybe it would be useful to install SP 4 on NT?!? I don't think that would help. It must be something else... Which arguments are you using at configure time? Do you have libz, libungif, libjpeg, libpng, XFree, etc? (I don't) -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Problems w. Compiling on WinNT
On Aug 11, 1999, [EMAIL PROTECTED] wrote: > [main] C:\bin\sh.exe 17910 (0) handle_exceptions: Exception: > STATUS_ACCESS_VIOLATION > [main] sh 17910 (0) handle_exceptions: Dumping stack trace to sh.exe.core Weird, I've just built it on Cygwin 20.1/WNT Server 4.0SP4 (maybe SP3?), on a 486!!!, and it appears to be working fine. Which versions (Cygwin, WNT) are you using? -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Compiling Kaffe 1.0b4 on Win96/CYGWIN/ make check fails
On Jul 13, 1999, Pavel Roskin <[EMAIL PROTECTED]> wrote: > I've attached a patch which fixes it. I couldn't find any patch, but I worked it out anyway. > Now HelloWorldApp can be compiled and run. However, its output contains > dos-style line ends, while the expected output (produced by sed) is in > unix style. > Thus, "cmp" finds the difference and the test fails. Yup. I found out filtering the program output through `grep .', that removes `\r's. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Still trying Kaffe/CYGWIN32--next problem: ";;" unexpected
On Jul 15, 1999, Phil Kelly <[EMAIL PROTECTED]> wrote: > It turned out that the CLASSPATH on the machine in question was set > and had a ; at the end of it! This was not the root of the problem. > (the eval line in the TestScript under the compile subroutine was > probably generating the error). Precisely. The classpath had to be quoted, to prevent semicolons from being interpreted as command separators. I'm installing a fix for this problem right now. > Now Kaffe.jar and pizza.jar apparently have syntax errors in them > when running make check (something about & unexpected and ) Because of the misinterpreted semicolons, it was trying to run the .jar files as if they were shell-scripts, which obviously doesn't work. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Class tree compilation "from zero"
On Aug 17, 1999, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > I'm not sure if pizza can handle this. (I don't know why it wouldn't > be able to, but I haven't tried it.) It can. That's exactly what we do in Kaffe. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: powerpc port
On Aug 13, 1999, jim hopper <[EMAIL PROTECTED]> wrote: > but if i look in HelloWorldApp.java at end where it gives expected > results and compare it with whats in HelloWorldApp.out expected > results match actual results, but for some reason it being shown as > failed. Run the testsuite with the VERBOSE environment variable set and it will explain what's going on. The *expected* output is saved in .out, while the actual, failed output is saved in .fail -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: configure problems
On Aug 11, 1999, jim hopper <[EMAIL PROTECTED]> wrote: >*.c | *.o | *.obj | *.xcoff ) ;; < added | *.xcoff I confirm that this is already fixed for autoconf 2.15. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: configure problems
On Aug 11, 1999, jim hopper <[EMAIL PROTECTED]> wrote: > checking for ld used by GCC... sed: -e expression #1, char 7: Unterminated > `s' command What do you get with `gcc -print-prog-name=ld'? What if you run configure with bash (CONFIG_SHELL=/bin/bash /bin/bash configure --args)? You might also try to get libtool.m4, ltconfig and ltmain.sh from the latest snapshot (or libtoolize -f with the latest libtool) and see if this changes anything... Or you can just set the environment variable LD to point to > /usr/lib/gcc-lib/powerpc-apple-machten4/2.8.1/ld > as to the size, i am doing a native compiler on machten using gcc > 2.8.1. Are there filename size limitations? For example, if the limit was 8 characters, it would try to read from `conftest' instead of `conftestval'. > configure:2429: gcc -o conftest.xcoff -g -fsigned-char conftest.c 1>&5 ^^ Ah, wait! This looks like an autoconf bug that mis-detects .xcoff as the executable extension, just because the linker creates a file with that extension *in addition* to the executable. I'm not sure the bug has already been fixed in newer releases of autoconf, but you can always override it by setting `ac_cv_exeext=""' in your environment. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Kaffe JIT for OSF Alpha
On Jul 29, 1999, Godmar Back <[EMAIL PROTECTED]> wrote: > sysdepCallMethod() should be the only significant hurdle here. sysdepCallMethod() is already working on alpha. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Kaffe JIT for OSF Alpha
On Jul 29, 1999, [EMAIL PROTECTED] (William E. Cohen) wrote: > When I force the configuration script to build with the JIT there > are a number of problem in the jit-alpha.def and the build is > unsuccessful. That's exactly the reason why config/alpha/osf/jit-md.h starts with: /* Needs update, do not use */ > Has anyone built the JIT for Alphas running OSF recently? Nope, it's been broken since the JIT2 revolution, that took place before Kaffe 0.10.0, if my memory serves. > Is anyone looking into this problem with the JIT not working for OSF > on the Alphas? It's been in my to-do list for months, but I don't think I'll be able to put much effort into it any time soon. :-( Contributions are definitely welcome. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Compiling Kaffe 1.0b4 on Win96/CYGWIN/ make check fails
On Jul 12, 1999, Phil Kelly <[EMAIL PROTECTED]> wrote: > But when I run 'make check' it fails with the following errors (top part of make >messages deleted for brevity): > ../../BUILD_ENVIRONMENT: /UnZip/kaffe-1.0b4/libraries/clib/net: not found I have fixed this problem on Kaffe just before releasing Guaraná 1.6 (based on June 1's post-1.0b4 snapshot of Kaffe), so you may just get the fix from CVS. Unfortunately, I haven't been able to get that particular snapshot of Kaffe running on cygwin :-( Has anybody had success with recent snapshots of Kaffe on cygwin? -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} ** I may forward mail about projects to mailing lists; please use them
Re: Kaffe's MS-Extensions Will Hurt Java
On Jun 24, 1999, Mike Cornall <[EMAIL PROTECTED]> wrote: > Where Java is concerned, Microsoft has stated that they intend to: > "kill cross-platform Java by grow[ing] the polluted Java market." > Tim is going to help them. Indeed :-( > Shame on you, Tim. I won't be using your product. You don't have to refrain from using the Standard part; you just have to fight against the non-Standard one. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Modifying prologue and epilogue
On Jun 20, 1999, Samarjeet S Tomar <[EMAIL PROTECTED]> wrote: > How can we modify prologue and epilogue to call a C method at the two > points to instrument the execution time of methods ? Try `grep -r prologue kaffe' :-) kaffevm/jit/icode.c and config/sparc/jit-sparc.def seem like good starting points. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: AWT : Automatic call to System.exit(0)
On Jun 19, 1999, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > *Browsing* 'Closed' might become difficult, but who really cares? I do. If we spend time classifying bug reports, it's just silly to throw that information away just because the bug was fixed. Category information should be retained somehow. > In my mind, the only reason to browse a category is to find the open > bugs in the category (e.g., to find something to fix). Isn't it easy enough to look for `unreplied' bug reports (assuming that `pending' ones already have an `owner')? > category, except its full of "closed" bugs. Picking out just the > open ones is impossible. Do I have to explain again how to look for `open' bug reports? > (Okay, currently Kaffe has 3 bugs in that category, so its really > not hard, but do you want the accumulated history of all network bug > reports to sit in that category for eternity?) Precisely. That's why there's a `net' category. Bug reports shouldn't change categories over time, otherwise I might not be able to find them by searching in the category I knew they lived in. >> ... since you can already know, in Jitterbug, whether a bug report >> is open or not: if it is `pending' or `unreplied', it is open, >> otherwise it is closed. > Creating a 'fixed' category and moving bugs to > there when they're really fixed seems like a simple and painless > solution. If you think classifying bug reports is useless, we could probably leave with just two categories: `open' and `closed'. What you don't get is that the classification is not just for the developers, when searching for open bug reports: users should also be able to search the categories to find out which bugs have been fixed and which have not. > Not more difficult than adding the note 'Fixed' that currently > litters the repository. In fact, a useless note, given that it's redundant. But it seems that I'm alone in the ``deep understanding'' of the underlying assumptions of Jitterbug. I wonder if Andrew Tridgell and Dan Shearer share this sensation with me... Or maybe it's I that just don't get it, and the `Message type's are just a useless concept in Jitterbug, that's going to be dropped in the next release... :-( -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: AWT : Automatic call to System.exit(0)
On Jun 18, 1999, Godmar Back <[EMAIL PROTECTED]> wrote: > We should add a "Closed" category to the bug database and > move closed bugs there. This way, you'll only see the > open bugs and don't get confused. That's exactly what we don't want to do, because we want people to be able to find closed bug reports too, so that we don't get multiple bug reports about the same problem, and so that they can tell whether upgrading to a newer version will fix their problem. If we created a single `closed' category, it would soon become a mess of unclassified bug reports, which would make it *harder* for people to search. This must be the 6th time I'll try to explain to Godmar and Archie the way Jitterbug works; it seems to me they just don't want to understand :-( I'm Cc'ing the jitterbug mailing list to get better coverage and more well-informed opinions. Unfortunately, Jitterbug doesn't offer a `closed' attribute. Its documentation suggests that `unreplied' and `pending' bug reports be considered open. The meaning of `unreplied' is quite obvious; `pending' is a bit harder, and that's where people get confused: a report is `pending' if it hasn't been ``touched'' after the last `followup' came in. ``touch'' means either `reply' or `change notes'. A `followup' is a message that comes from the user (or anybody else, for that matter, with a subject that identifies the Problem Report), with additional information on that bug report. So it's not a matter of being complicated (it is not!), it's just a matter of understanding the concepts of the tool we're trying to use. If it was done the way it was, there's a reason for that. You have three choices: (i) understanding the underlying assumptions of the tool and using it as it was supposed to be used, (ii) introducing complication to make it work in a way it was not designed to, or (iii) adopting a tool that matches your expectations. Jitterbug assumes that, when you reply to a bug report, you're saying `thanks for your bug report. The bug was fixed on release 1.2.3'', or asking for further details, in which case you expect a followup. When such a followup comes in, the bug report becomes `pending'. But maybe it was just the user saying `thank you', so you may not want to reply once again, then you can close it by just `changing notes'. We could create one `*-closed' category for each `*' category, but that means that, whenever we `reply' to a bug report, closing it, we'd *also* have to move it from one category to another. Moreover, we'd be introducing redundancy, since you can already know, in Jitterbug, whether a bug report is open or not: if it is `pending' or `unreplied', it is open, otherwise it is closed. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Calling native methods in Java
On Jun 16, 1999, Samarjeet S Tomar <[EMAIL PROTECTED]> wrote: > I want to call a native method from Java code. I have written the code, > compiled it and created the library also. But when I run it with Kaffe it > gives the following error : > Failed to locate native function: > spec/harness/Mytime.st_time()V > java.lang.UnsatisfiedLinkError Which platform? How did you implement the native method, with JNI? How did you load the native class? -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: AWT : Automatic call to System.exit(0)
On Jun 17, 1999, Peter Mehlitz <[EMAIL PROTECTED]> wrote: > Rodrigo Senra wrote: >> I dug a little and found out at class Default a static field AutoStop > Right, that's how it is controlled. The comments for that field just tend to > vanish each time I document it again. Set it to "false" and you should have > the "standard" behavior. I'd rather have the `standard' behavior, in this case. You can always ^C if it doesn't terminate, but you can't get it to proceed from where it stopped if it just exit()s :-( I'd vote for having this defined in a property file, or as a command-line property (say, -Dkaffe.awt.AutoExit=true) -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Please point out my stupidity! (Newbie question)
On Jun 12, 1999, Gerard Hickey <[EMAIL PROTECTED]> wrote: > %> java Hello.class Try `java Hello' -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Java profiler for kaffe
On Jun 8, 1999, Godmar Back <[EMAIL PROTECTED]> wrote: > For these, the obvious solution is to not jit them (a la hotspot) --- > but we need the infrastructure for that in place! Somebody (check the > mailing list archives please) did this already, but like so many > things it was either never released or never made it back into the > distribution (I think the former). Actually, the guy (whose name I also fail to recall) offered to post the code, but, when I asked him to send it to me, so that I could make it available in the Kaffe CVS tree or something, I never got a reply back (but take this history with a grain of salt, my memory plays tricks on me these days :-) The code was i386-specific. I don't know how hard it would be to port it to other architectures. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Question about errormessages!
On Jun 1, 1999, "Maria Kling" <[EMAIL PROTECTED]> wrote: > Kaffe: code-analyse.c:97: verifyMethod: Assertion `codeInfo == 0 && " > Attempt to reenter verifier!"' failed. > Abort (core dumped) You probably need a newer release of Kaffe. I believe this problem is already fixed in 1.0.b4, but getting the latest snapshot wouldn't hurt :-) -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: libtool-enhanced testing on HP-PA
On May 28, 1999, Kero van Gelder <[EMAIL PROTECTED]> wrote: > A miracle just happened: compilation out-of-the-box of Kaffe managed > to launch the biggest Java package I have written. > That is, with shared libraries! (Somebody else may try the static ones :-) > Kaffe works under HP/UX 10.20! This is very good news! But to give credit where credit is due, the ``gods'' responsible for the ``miracle'' :-) are John David Anglin and David Jones, for having fixed libltdl support for HP/UX, and Gary V. Vaughan, from the Libtool team, that spent a lot of time with me investigating how to properly create shared libraries on HP/UX. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Newbie question
On May 27, 1999, Gunnar Stahl <[EMAIL PROTECTED]> wrote: > Today I've downloaded and installed the kaffe 1.0b4 on my machine. > During installation I've run the "make check" tests and thist told me > that everything is ok. Not really. The testsuite is quite limited; it doesn't test anything of AWT, for example. > Then I tried to run the IDE I`m working with for a year now (netbeans > developer 2.1.1, completely written in java using jdk1.1 and jfc1.1). > When I start it after a few seconds I get the error > "java.lang.ExceptionInInitializerError [exception was > java.lang.NullPointerException]". You may want to try a CVS snapshot; several bugs have been fixed since release 1.0.b4 -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: String.indexOf(String, int) bug
On May 26, 1999, Colin Stevens <[EMAIL PROTECTED]> wrote: > String.indexOf(String str, int offset) crashes if str is a 0-length string, Thanks, fixed. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: jitmem stat fix
On May 21, 1999, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > The 'jitmem-codeblock' stat was being updated incorrectly. This patch > to kaffevm/jit/machine.c fixes it. Thanks, installed (I thought you had write permission, but... :-) -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: To change a method
On May 21, 1999, Ivana Visconti Menna <[EMAIL PROTECTED]> wrote: > I want to change, for example, the method writeObject of class > ObjectOutputStream, but I found several methods. > What is the diference between > kaffe-1.0b4/libraries/javalib/java/io/ and > kaffe-1.0b4/libraries/javalib/kaffe/io ? > What method must I change? It depends on what you want to do. java/io/ObjectOutputStream is mostly a wrapper for kaffe/io/ObjectOutputStreamImpl. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: vmdebug uses both stdout and stderr (PR#102)
On May 22, 1999, Pavel Roskin <[EMAIL PROTECTED]> wrote: >> The code should only use dprintf. It's up to debug.c >> to decide where the debugging output actually goes. >> (Normally stderr) > + * kaffe/kaffevm/*, libraries/clib/zip/Deflater.c: printf replaced > + with dprintf inside debug macros. Thanks, installed -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: diffs for time and debugging
On May 17, 1999, Godmar Back <[EMAIL PROTECTED]> wrote: > I guess it would help if you pointed us at where libtool says this > information when you remove that patch, checking whether to build shared libraries... yes checking whether to build static libraries... no -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: diffs for time and debugging
On May 13, 1999, Patrick Tullmann <[EMAIL PROTECTED]> wrote: > * configure.in: be verbose about static/shared libraries, Even though Archie has already installed this patch, I'm inclined to revert it, because it will report incorrect information on platforms that do not support shared libraries. ltconfig already reports whether shared or static libraries are going to be created, and then, it can be 100% positive, which your test cannot. Would you mind if I reverted it? -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Kaffe when blackdown in on.
On May 1, 1999, Alex Turner <[EMAIL PROTECTED]> wrote: > On 29 Apr 1999, Alexandre Oliva wrote: >> On Apr 28, 1999, Alex Turner <[EMAIL PROTECTED]> wrote: >> >> > I am having some trouble under RH5.2. Once I've put blackdown on the >> > system, kaffe refuses to run, and simply segfaults. Even the compiler >> > wont go! >> Check your CLASSPATH. > How exactly does that help me? > Kaffe is executing, but it just segfaults. How does my CLASSPATH affect > this? Different CLASSPATHs may result in different offsets for fields of some low-level system classes. Kaffe pre-1.0.b4 wouldn't check whether it's using the appropriate bootstrap classes, and would simply crash. Newer releases do check that, and try to flag an incorrect CLASSPATH. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Kaffe Cannot run in SunOS5.7
On May 5, 1999, John Pierce <[EMAIL PROTECTED]> wrote: > I don't know about 1.0b4 but the current CVS tree builds and runs with a > minor libtool fix. What libtool fix? > I was unable to get it to pass the test suite with any level of > optimization enabled w/gcc 2.8.1 or egcs-1.1.2. W/o -O everything > seems to work fine. Weird... Last time I tried (maybe 1 or 2 weeks ago), it passed all tests on Solaris 7, both /sparc and /x86, with egcs 1.1.2. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: libtool-enhanced testing on HP-PA
On May 13, 1999, David Jones <[EMAIL PROTECTED]> wrote: > I tried to build Kaffe on HPUX 10.20 on a PA8000 box. > I get errors when I try to run it - the shared libraries are not found. Which shared library isn't found? Is it running in the build dir or after install? Which version/snapshot is that? There were some recent libtool tweaks for HPsUX (it really does! :-(, particularly for the dlopening mechanism, after release 1.0.b4, so a snapshot may be worth a try. > First of all, LD_LIBRARY_PATH does not so anything; you need to set > SHLIB_PATH on HP-UX. Yep, but it doens't matter. Kaffe doesn't really look at LD_LIBRARY_PATH or SHLIB_PATH; it's only KAFFELIBRARYPATH that matters. You'd still be able to arrange for libkaffevm (the only library linked into Kaffe) to be found by tweaking LD_LIBRARY_PATH or SHLIB_PATH, but SHLIB_PATH is not searched by default on HPsUX. It would only be searched if the +s switch was used when creating the executable, but libtool doesn't use this switch, in order not the change the system's default. Since it hard-codes the correct search paths into the executable, this shouldn't be a problem. > Also, libraries end in .sl instead of .so. The objects created do > indeed end in .sl, but I'm not convinced that the system is looking > for them properly. It's supposed to be. If you don't think so, check libltdl/config.h; it should `#define LTDL_SHLIB_EXT ".sl"' > Since HP-UX has neither working system call tracing (at least it doesn't work > too well on our installation at work) or "ldd" AFAIR, HPsUX's ldd is called chatr (sp?). > I am at a loss as to how to continue, other than to build static. If this worked, it would already be a good start :-) > Has the HP version been tested since the integration of libtool? Yep, but not with success so far :-( Only two weeks ago someone spotted and fixed a bug in libltdl that would prevent libltdl from minimally working on HP/UX. It's reported to be working now (in libtool 1.3), but I don't think anyone has tried Kaffe snapshots after my last libtool import. I'd appreciate any feedback. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Kaffe & Apache
On May 11, 1999, Michael Olan <[EMAIL PROTECTED]> wrote: > I'm using a really old kaffe-0.7.1 on m68k under NextStep3.3. > What's the latest version that works on this system? Kaffe seems to be working on m68k again, after a lot of work by Kiyo Inaba and David Jones. At least on Linux and NetBSD. It's probably worth a try on NextStep too. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Help Compiling under Solaris
On May 11, 1999, Kris Bubendorfer <[EMAIL PROTECTED]> wrote: > I've tried compiling on a Sparc Solaris with little success. Configure > completes, but seems to generate a bad MakeFile, Actually, it seems that libtool configuration isn't capable of figuring out a command to extract symbol lists from libraries, which is quite weird, because it works for me on several releases of Solaris/sparc, with various compilers and linkers :-( Which release or snapshot of Kaffe are you using? What do you get by running: ./libtool --config | grep '^global_symbol_pipe=' What are the lines just after `checking if global_symbol_pipe works' in config.log? Thanks -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Kaffe when blackdown in on.
On Apr 28, 1999, Alex Turner <[EMAIL PROTECTED]> wrote: > I am having some trouble under RH5.2. Once I've put blackdown on the > system, kaffe refuses to run, and simply segfaults. Even the compiler > wont go! Check your CLASSPATH. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Is it possible for Kaffe to be interoperatable with Sun's in serialization?
On Apr 28, 1999, "J.H. Liu" <[EMAIL PROTECTED]> wrote: > Is it possible for Kaffe to be interoperatable with Sun's JVM in > serialization? Yup. Sun has recently released a lot of information about the standard representation of serializable Core classes, which made it possible for us to be compliant, even if the internal implementation and representation of classes is different. We may have to write explicit serialization code, though. TVT has recently contributed serialization code for a lot of classes, but I'm not sure they've implemented them all already. > Is it possible to serialize & deserialize StringBuffer object across > Kaffe's and Sun's JVM. You'd have to test it. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Libtool problems with Kaffe B4 and Cygnus WIN32
On Apr 26, 1999, Pavel Roskin <[EMAIL PROTECTED]> wrote: >> It falls down for me when it tries to load the sysv shm and ipc >> headers (sys/shm.h and sys/ipc.h) which don't exist on vanilla >> cygwin =(O| > Patch attached. Thanks. I installed a slightly modified version of it: >AC_CHECK_LIBRARY(Xext,XShmQueryExtension,X_PRE_LIBS) This will define HAVE_LIBXEXT if it finds the function, so we don't need this: > + if test x"$ac_cv_lib_Xext_XShmQueryExtension" = x"yes"; then > +AC_DEFINE(HAVE_SHM, 1, [Define to support X shared memory extension]) > + fi after replacing HAVE_SHM with HAVE_LIBXEXT in the rest of your patch. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Unneeded acinclude.m4 in Kaffe
On Apr 26, 1999, Pavel Roskin <[EMAIL PROTECTED]> wrote: > I just wonder what's the reason to have acinclude.m4 and > libltdl/acinclude.m4 in Kaffe? They're copies of libtool's libtool.m4, and their existence ensures that an incompatible, pre-installed version of libtool.m4 won't be used in any case. > It is even possible that those macros are not up-to-date Nope, I update them whenever I update the rest of the libtool utilities. > What about acinclude.m4, it should only contain definition for > AC_CHECK_LIBRARY, which now resides in configure.in Yup, moving it to acinclude.m4 would be nice, but there's no pressing need for it, anyway. Its definition closer to its use is somewhat clearer for autoconf newbies. And then, aclocal is deemed to die, we just haven't figured out a good design to be able to drop it. So I'd rather keep things as they are. > By the way, AC_CHECK_LIBRARY is a bad name. I tried to find it in Autoconf > documentation. kaffe_CHECK_LIB would be more appropriate. Maybe AC_kaffe_CHECK_LIB. Defining macros started with AC_ is good because autoconf will warn if any such macro is not expanded for some reason. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: some small patches
On Apr 25, 1999, Archie Cobbs <[EMAIL PROTECTED]> wrote: > Alexandre Oliva writes: >> > What documentation are you looking at? This one: >> >> > http://java.sun.com/products/jdk/1.2/docs/api/java/io/RandomAccessFile.html >> >> > says to throw FileNotFoundException for the (String, String) >> > constructor. >> >> Yup, but it says that the method is *declared* to throw IOException. > My eyes read otherwise.. are you sure you're not looking at the > (File, String) constructor instead, which does throw IOException? Damn! No, I'm not sure. In fact, I was looking at the wrong constructor :-( Sometimes I wish Java did not support method/constructor overloading :-) Will you please install this missing part of Aaron's patch? Sorry about all the trouble :-( -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: some small patches
On Apr 25, 1999, Archie Cobbs <[EMAIL PROTECTED]> wrote: > Alexandre Oliva writes: >> >* libraries/clib/io/RandomAccessFile.c, >> >libraries/javalib/java/io/RandomAccessFile.java: change >> >constructor to throw FileNotFoundException >> >> Throwing FileNotFoundException is correct, but the method declaration >> still says it throws IOException, so I did not install the .java patch. > What documentation are you looking at? This one: > http://java.sun.com/products/jdk/1.2/docs/api/java/io/RandomAccessFile.html > says to throw FileNotFoundException for the (String, String) > constructor. Yup, but it says that the method is *declared* to throw IOException. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: [PATCH] Use Jikes as default compiler if installed (autoconf version.)
On Apr 25, 1999, Niels Kristian Bech Jensen <[EMAIL PROTECTED]> wrote: > This patch makes kaffe use Jikes as its default compiler rather than > pizza if Jikes is installed when "configure" is executed: Wouldn't it be better to have a separate `javac' script for this case, in a separate directory, that would set JIKESPATH just like `kaffe' sets CLASSPATH and run jikes directly? I, for one, would find it surprising if javac finding servlet classes depended on jikes not being installed... -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: some small patches
On Apr 18, 1999, Aaron Place <[EMAIL PROTECTED]> wrote: > Could someone please commit these changes (if they are correct). Thanks, installed most of it. > * libraries/clib/io/RandomAccessFile.c, > libraries/javalib/java/io/RandomAccessFile.java: change > constructor to throw FileNotFoundException Throwing FileNotFoundException is correct, but the method declaration still says it throws IOException, so I did not install the .java patch. > * libraries/javalib/java/io/StringReader.java (ready): fix Doesn't make much sense to me; `closed' would never be `true'. Maybe you meant to set it to `true' in close(). And then, shouldn't it be checked in *every* method, not only in ready()? > * libraries/javalib/java/io/StringWriter.java (StringWriter): make > public > (close): change signature Nope, close() is documented as throwing IOException, we shouldn't drop it. But maybe it shouldn't do nothing, behaving like in your proposed change to StringReader. > * libraries/javalib/java/text/DateFormat.java, NumberFormat.java: > clone is handled by superclass The docs say they *do* define clone(), they just don't explicitly implement Cloneable. I've installed only the part that removes Cloneable from `implements'. Thanks! -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: 5 misc fixes
On Apr 20, 1999, Robert Zawiasa <[EMAIL PROTECTED]> wrote: > fix#1 in kaffe/libraries/clib/awt/X/fnt.c > fix#2 in kaffe/libraries/javalib/java/applet/Applet.java > fix#3 in kaffe/libraries/javalib/java/beans/PropertyChangeEvent.java and > kaffe/libraries/javalib/java/beans/PropertyVetoException.java > fix#4 in kaffe/libraries/javalib/java/lang/reflect/Member.java Thanks, installed these too. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: FreeBSD linking problem
On Apr 23, 1999, Godmar Back <[EMAIL PROTECTED]> wrote: > Why not add them to the end, so that standard libs installed > in /usr/local/lib are still found, but the newly built libs > will still override any garbage you may have in /usr/local/lib? Yup. Adding them to LIBS, not to LDFLAGS, is the right approach. However, since AC_CHECK_LIB *pre*pends to LIBS, we might have trouble with dumb linkers that search for a library only in -Ldirs specified *before* the library. But I don't know of any such linker, so we can probably live with it, and I'm installing a patch shortly :-) -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: java.awt.Choice silliness
On Apr 22, 1999, "John B. Lee" <[EMAIL PROTECTED]> wrote: > Has anyone experienced problems whereby selecting > a choice from a java.awt.Choice component produces > the follow exception: > java.lang.NullPointerException > at java.awt.RowCanvas.repaintRows(RowCanvas.java:257) A patch that may fix this problem was installed yesterday. Note that Klasses.jar was not rebuilt, though, so you'd have to `make build-classes' to update it. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: FreeBSD linking problem
On Apr 22, 1999, David Jones <[EMAIL PROTECTED]> wrote: > The problem? Kaffe is linked using -lkaffevm, and libkaffevm is dynamic. > So, ld searched my LD_LIBRARY_PATH for it. The old library, installed in > /usr/local/lib, was hit first, and ld couldn't find the symbols. Choke. Nope, it has nothing to do with LD_LIBRARY_PATH or libkaffevm being dynamic or not. The problem is that kaffe's configure.in automatically adds -L/usr/local/lib and -I/usr/local/include to LDFLAGS and CPPFLAGS, respectively, in a way that they're searched before files living in the source or build tree. So pre-installed versions of libraries and headers may be used instead of newer ones. IMO, that's a good argument for not adding those directories to the standard search paths unless the user explicitly says so. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Algorithm used by BigInteger prime generator?
On Apr 22, 1999, "John Keiser" <[EMAIL PROTECTED]> wrote: > Incidentally, as long as this method creates random BigIntegers, why does it > matter what algorithm is used? Given my Random generator, initialized with the same seed, I should be able to generate the same sequence of prime BigIntegers on any Java platform, otherwise WORA falls apart. That's why the standard Random class is clearly specified, and that's why the algorithm for random big prime generation should too. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Algorithm used by BigInteger prime generator?
On Apr 22, 1999, Andrew Haley <[EMAIL PROTECTED]> wrote: >> Except that this may generate a prime with more than bitLength bits, > Did you not see my comment that "a real version should check for > overflows"? Yup, I should have ack'ed it. >> Furthermore, it doesn't generate randomly distributed primes > The specification doesn't require it to. Indeed. It doesn't really say much. > The technique of simply adding 2 until a prime is reached is pretty > much standard in crypto, and with crypto-sized BigIntegers the > deviation from a uniform distribution is insignificantly small. But we're not talking about crypto-sized BigIntegers, we're talking about the BigInteger class. It's supposed to be general-purpose, not exclusive for use for cryptography. If it were, it probably wouldn't be exportable :-) > If this is really unacceptable, a new random integer could be > generated with each iteration, which answers such objections, but > it's an extra run-time overhead. But then it certainly won't meet the requirement about execution time proportional to the value of certainty. >> and I'm not sure this meets the requirement of O(certainty). > I don't understand this. Are you saying that isProbablePrime > (certainty) does not meet the requirement of O(certainty)? Nope, I'm just unsure whether testing primes sequentially does, assuming that isProbablePrime takes constant time for a constant bitLength. >> Nevertheless, the specification leaves room for a lot of different >> implementations, particularly in the use of rnd, which is unacceptable >> for WORA. > I see your point. However, the requirement that all implementations > return the same result means that implementors are prevented from > inventing a more efficient prime generator. In real-world crypto > applications, where random primes are frequently generated, this is > very bad news. In financial applications, it is not uncommon to use > hardware accelerators to generate random primes: it would be > impossible to use such hardware if the exact algorithm used to > generate primes was fixed. You're free to implement whatever algorithm you want for big prime generation in your own classes, but java.lang.BigInteger.BigInteger(int,int,Random) must be completely specified in order to allow for WORA. If this means using an inefficient implementation, so be it. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Algorithm used by BigInteger prime generator?
On Apr 22, 1999, Andrew Haley <[EMAIL PROTECTED]> wrote: > As far as I can see, what is intended is something like this: > if (candidate.isProbablePrime (certainty)) > return candidate; > else > candidate = candidate.add (BigInteger.valueOf (2)); Except that this may generate a prime with more than bitLength bits, depending on the initial random number and on the number of failures you get, so you may have to retry if it overflows. Furthermore, it doesn't generate randomly distributed primes, and I'm not sure this meets the requirement of O(certainty). Anyway, I've just committed a version that does it. But it could be made much more efficient by using a sieve, since isProbablePrime is quite expensive. Nevertheless, the specification leaves room for a lot of different implementations, particularly in the use of rnd, which is unacceptable for WORA. Theoretically, given the same implementation of Random, with the same seed, one should get the same sequence of random primes, regardless of the JVM used. Also note that your implementation will never generate 2 for bitLength==2. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Libtool problems with Kaffe B4 and Cygnus WIN32
On Apr 22, 1999, Pavel Roskin <[EMAIL PROTECTED]> wrote: > This is clearly a bug in libtool (maybe triggered by some uncommon things > in Kaffe's configure.in) Yup. It seems that it does not properly support --disable-static on cygwin :-( OTOH, it won't create DLLs yet, so you might well go with --disable-static, and everything should work. I hope this can be fixed in libtool as soon as cvs.gnu.org comes back to life. It's been down for more than a week already :-( -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: local_check_header
On Apr 21, 1999, John Song <[EMAIL PROTECTED]> wrote: > java.lang.ExceptionInInitializerError: [exception was >java.lang.UnsatisfiedLinkError: Failed to locate native function: >java/math/BigInteger.initialize0()V] > at Big.main(Big.java:5) Make sure you have GNU gmp installed, and that `configure' has found it. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Algorithm used by BigInteger prime generator?
On Apr 20, 1999, Paul Fisher <[EMAIL PROTECTED]> wrote: > Alexandre Oliva <[EMAIL PROTECTED]> writes: >> Does anybody know what algorithm the constructor >> java.math.BigInteger.BigInteger(int bitLength, int certainty, Random >> rnd) is supposed to use? > Sun's java.math implementation uses Colin Plumb's <[EMAIL PROTECTED]> > BigNum library ftp://skip.incog.com/pub/bnlib-1.1.tar.gz>. > You'll find the prime number generation routines there. The problem is that there's no specification about what to do after generating a random number, nor does it talk about ``adjusting'' a randomly-generated number so that it passes a primality test. For example, Plumb's primeGen doesn't provide a mechanism to specify a `certainty'. Furthermore, Plumb's implementation accepts a rand function as an argument, that takes only an unsigned integer as argument, to randomize the increments to the base random number. I don't see how Sun might have written a wrapper function that would call back the Random generator passed as argument, since it wouldn't be able to figure out which Random object to invoke... But then, this is not an issue, since we wouldn't be using Plumb's implementation anyway. I'm going to install a very dumb implementation, and improve it when I get input from Sun about the actual algorithm that should be used. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Algorithm used by BigInteger prime generator?
On Apr 20, 1999, Paul Fisher <[EMAIL PROTECTED]> wrote: > Alexandre Oliva <[EMAIL PROTECTED]> writes: >> Does anybody know what algorithm the constructor >> java.math.BigInteger.BigInteger(int bitLength, int certainty, Random >> rnd) is supposed to use? > Sun's java.math implementation uses Colin Plumb's <[EMAIL PROTECTED]> > BigNum library ftp://skip.incog.com/pub/bnlib-1.1.tar.gz>. > You'll find the prime number generation routines there. Thanks for the pointer. BTW, this is GPLed, did Sun get a special license from Colin Plumb? > Implementing the algorithm for Classpath is on my TODO list. > Although, if you're interested in writing it, we'd love to include > it. :) Well, I'm currently writing it for Kaffe, but if it fits well into Classpath, or requires a just little adaptation, I'd be glad to contribute it to Classpath too. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: Algorithm used by BigInteger prime generator?
On Apr 21, 1999, Godmar Back <[EMAIL PROTECTED]> wrote: > I'm wondering whether you have to use BigNum. I hope not. > Specifically, I was wondering if some of the invariants concerning > the representation of the magnitude and the location of the MSB are > the same. I plan to look into it some day, since I want to get MindTerm (a GPLed ssh client) to work with Kaffe. Anyway, I've filed a Java specification bug report, and they acknowleged it as a bug. It should soon appear in the Bug Parade as Bug Report # 4231449. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists
Re: no more gif problem + nice surprise
On Apr 20, 1999, Peter Mehlitz <[EMAIL PROTECTED]> wrote: > Alexandre Oliva wrote: >> it's the appropriate version. Then we'll be able to drop the >> header-file-check hack, which we can't always trust anyway. If >> there's no new function in libungif.so.4, we could perform the > Definitely. The right function would be DGifOpen, which is the > specific thing we need from the new libungif (the DGifOpenFile isn't > used, anymore). Ok, this should be fixed in the CVS tree in a few minutes. The check for the appropriate header file will still be performed, though, to ensure that we don't get weird compile errors if the library path is properly set to find the newer version, but the include path is not. -- Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Brasil {oliva,Alexandre.Oliva}@dcc.unicamp.br aoliva@{acm.org,computer.org} oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org} *** E-mail about software projects will be forwarded to mailing lists