[Fluxus] Trouble with FluxusInMred OSX - segfault in load-extension

Scott alcoholiday at gmail.com
Wed Mar 4 13:39:55 PST 2009


Howdy!
First - Fluxus is awesome, and I'd like to use it for a big project.
It took me a while, but I managed to get Fluxus to build on my 10.5.6
MacBook... and I plan on documenting what I ended up doing in another post
(one tip: "sudo scons darwin=1 PLTLib=/usr/local/lib /" <- note trailing
slash!). Never used scons  before, which was part of the issue...
regardless!

I built PLT scheme (drscheme, mred, etc) from source and stuck it in
/usr/local, and it works like it should.
I built Fluxus and it works (Fluxus.app) just swell too!

However, when I try the following in DrScheme:
(load-extension
"/usr/local/collects/fluxus-015/compiled/native/i386-macosx/3m/fluxus-engine_ss.dylib")


I get a segfault in the fluxus-engine scheme_reload function, specifically
it blows up doing scheme_intern_symbol since the symbol_table pointer is 0.

(gdb) backtrace
#0  symbol_bucket (table=0x0, key=0x1e1c76d0 "fluxus-engine", length=13,
naya=0x0) at symbol.c:113
#1  0x1e5f608c in scheme_intern_exact_symbol_in_table_worker
(symbol_table=0x0, kind=0, name=0x1e1c76d0 "fluxus-engine", len=13) at
symbol.c:377
#2  0x1e5f6141 in scheme_intern_exact_symbol (name=0x0, len=0) at
symbol.c:419
#3  0x1e5f644f in scheme_intern_symbol (name=0x1e1c76d0 "fluxus-engine") at
symbol.c:491
#4  0x1e142caa in scheme_reload (env=0xd4bf88) at
modules/fluxus-engine/src/FluxusEngine.cpp:313
#5  0x007ddd4a in scheme_default_load_extension (argc=2, argv=0xbfffd5e4) at
dynext.c:462

At this point, I've decided to call in the experts.

One thing I was wondering... are there two symbol_table s? One defined in
the MrEd shared library and another in the PLT_MzScheme that is loaded by
the fluxus-engine.

Thanks for all your help, obviously, let me know what else I can send.

-Scott
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.pawfal.org/pipermail/fluxus-pawfal.org/attachments/20090304/3f519e92/attachment-0001.htm>


More information about the Fluxus mailing list