Hi all, August,
I'm getting lots of segfaults from readanysf 0.13 with PD 0.38-2 from CVS stable with Guenter's help browser patch.
I put it through gdb and got the following. Note the "invalid command name..." seems to always be there and has no apparent connection to readanysf~.
I get an error when I try to stop playing an ogg file with [stop(
<gdb> Starting program: /usr/local/bin/pd [Thread debugging using libthread_db enabled] [New Thread -1210414336 (LWP 8346)] [New Thread -1210942544 (LWP 8351)] PDP: version 0.12.2 [New Thread -1225798736 (LWP 8352)] [New Thread -1235555408 (LWP 8353)] [New Thread -1410163792 (LWP 8354)] [New Thread -1418552400 (LWP 8355)] [New Thread -1427735632 (LWP 8356)] invalid command name ".x83deee8.c" invalid command name ".x83deee8.c" ReadVorbis:: calling close function
Program received signal SIGSEGV, Segmentation fault. [Switching to Thread -1427735632 (LWP 8356)] 0xb7e1aa35 in free () from /lib/tls/libc.so.6 (gdb) run
</gdb>
And when I try to play an mp3 I get this:
<gdb> [New Thread -1427735632 (LWP 8415)] invalid command name ".x83deee8.c" invalid command name ".x83deee8.c" MAD_ERROR_BUFLEN
Program received signal SIGSEGV, Segmentation fault. [Switching to Thread -1427735632 (LWP 8415)] 0xb7e1bb12 in mallopt () from /lib/tls/libc.so.6 (gdb) run
</gdb>
or this:
<gdb> [New Thread -1427735632 (LWP 8424)] invalid command name ".x83deee8.c" invalid command name ".x83deee8.c" ReadMad::couldn't read from InputFIFO, bailing...
Program received signal SIGSEGV, Segmentation fault. [Switching to Thread -1427735632 (LWP 8424)] 0xb7e1bb12 in mallopt () from /lib/tls/libc.so.6 (gdb)
</gdb>
Any thoughts? Oh, I'm running this on Debian SID w/ libmad 0.15.1b-1 and libvorbis 1.0.1-1
Best, ian
Any thoughts? Oh, I'm running this on Debian SID w/ libmad 0.15.1b-1 and libvorbis 1.0.1-1
what version of FLEXT are you driving?
any version greater than 0.4.5 will crash PD hard. this has to due with the object wrapping in newer FLEXT versions.
Flext > 0.4.5 will wrap both "new" and "delete" calls in c++ to use pd's internal byte allocation/deallocation routines.
sorry about this. readanysf~ works fine with flext 0.4.5. Not sure why Thomas and other flext users are able to do "new" and "delete" calls in their code. I can't seem to nail down why that don't jive with my code.
-august.
Jeez, I'm sorry. I had actually seen that in the archives and decided to set readanysf~ aside for the time being. When I came back to it I forgot that I already knew what was causing the problem ;) That's a shame since a bunch of other externals (at least the versions I'm using) require flext 0.4.6.
Ummm... I wish I could debug it, but my knowledge of flext is nonexistant. Maybe something to do with the new build system? (which is very handy, btw).
Cheers, and sorry to disturb,
-Ian
august wrote:
Any thoughts? Oh, I'm running this on Debian SID w/ libmad 0.15.1b-1 and libvorbis 1.0.1-1
what version of FLEXT are you driving?
any version greater than 0.4.5 will crash PD hard. this has to due with the object wrapping in newer FLEXT versions.
Flext > 0.4.5 will wrap both "new" and "delete" calls in c++ to use pd's internal byte allocation/deallocation routines.
sorry about this. readanysf~ works fine with flext 0.4.5. Not sure why Thomas and other flext users are able to do "new" and "delete" calls in their code. I can't seem to nail down why that don't jive with my code.
-august.