Re: SWT on darwin

["Geoff Strom" <stromgt at speakeasy dot net>]

> Andreas Tobler wrote:
> 
> Geoff Strom wrote:
> > LiveConnect in Rhino now works again (gcc-3.4-20040317) on darwin.
> > In fact things look pretty good for darwin and 3.4.
> > 
> > But, Eclipse SWT still doesn't seem to want to work. It tries, but dies. This time
> > with a Segmentation fault instead of a Bus error. I've attached a crash
> > log if it helps any.
> 
> Hm, maybe Tom or others can comment if Eclipse is running with recent 
> gcc's. Never tried.
> 
> > 
> > Are there any outstanding patches for darwin?
> 
> I assume you added the things I mentionend last time?
> 
> If so, I don't know of anything which is missing.
> 
> Well, I can't follow the development since my main machine died and I 
> can not test on 10.3.3.
> 
> Would you mind to try a gdb dump of the TabFolderTest app?
> 
> A darwin trace does not tell me that much.
> 
> #gdb TabFolderTest
> gdb>r	(run the app)
> gdb>bt full (after the SEGFAULT occured do a backtrace)
> 
> This may give some more insight.
> 
> Andreas

I guess I'll have to check if the two patches you mentioned
were in fact in the snapshot I tried (gcc-3.4-20040317).

I stepped through TabFolderTest with gdb and may have
found something that will help. Attached is a small snippet
of the gdb session which precedes the crash.

It looks like TabFolderTest is crashing when attempting to
get or load the address of the first native function it tries
to use (which is in the SWT JNI dylib). Of course this seems
strange, because JNI seems to work otherwise.

Two interesting things in the gdb session stand out:

_Jv_FindSymbolInExecutable(char const*)
(symname=0x1 <Address 0x1 out of bounds>)
at ../../../gcc-3.4-20040317/libjava/java/lang/natRuntime.cc:85

and later,

throw new java::lang::IllegalMonitorStateException;

Thanks,
Geoff

Attachment: tabfoldertest_gdb_session.txt
Description: Binary data