Re: [PD] Crasher bug
More useful is to print the frame within the pd code where the crash occurred. In this case, try the following in lldb: f 6 f That should print the section of code with the crashing function and/or variable underlined. The stuff below in the kernel is no so useful to see. > On Oct 2, 2017, at 5:49 PM, Roman Haefeliwrote: > > On Mon, 2017-10-02 at 13:14 +0200, Dan Wilcox wrote: >> If you want more info, you can compile pd with debugging symbols >> added via: >> >> ./configure --enable-debug CFLAGS="-g" >> >> I'm going to add the "-g" back to the default debug settings as it's >> needed for useful stack frame info, etc. >> >> If you have the Xcode command line tools installed (not the Xcode >> app!), you can use lldb for similar info: http://lldb.llvm.org/lldb- >> gdb.html. It can give you the exact location in the code where the >> crash occurred. > > Thanks, Dan. > > Just in case this is useful to anyone, here is the backtrace on macOS: > > $ lldb -- pd -stderr -rt -jack -jackname tpf-client-1 -nojackconnect > -inchannels 8 -outchannels 64 -open > /Users/itz/pd-src/tpf/pd-tools/tpf-client.pd > (lldb) target create "pd" > Current executable set to 'pd' (x86_64). > (lldb) settings set -- target.run-args "-stderr" "-rt" "-jack" "-jackname" > "tpf-client-1" "-nojackconnect" "-inchannels" "8" "-outchannels" "64" "-open" > "/Users/itz/pd-src/tpf/pd-tools/tpf-client.pd" > (lldb) run > Process 28469 launched: '/usr/local/bin/pd' (x86_64) > Pt_Start() called > slipenc: maximum packet length is 16384 > iemnet - networking with Pd: [tcpclient] > compiled on Dec 14 2015 at 14:29:01 > copyright © 2010-2015 IOhannes m zmoelnig, IEM > based on mrpeach/net, based on maxlib > slipdec: maximum packet length is 16384 > iemnet - networking with Pd: [udpclient] > compiled on Dec 14 2015 at 14:29:01 > copyright © 2010-2015 IOhannes m zmoelnig, IEM > based on mrpeach/net, based on maxlib > pd(28469,0x7fff79118000) malloc: *** error for object 0x1020f0c08: incorrect > checksum for freed object - object was probably modified after being freed. > *** set a breakpoint in malloc_error_break to debug > Process 28469 stopped > * thread #1: tid = 0x1c38a, 0x7fff97b02f06 > libsystem_kernel.dylib`__pthread_kill + 10, queue = 'com.apple.main-thread', > stop reason = signal SIGABRT > frame #0: 0x7fff97b02f06 libsystem_kernel.dylib`__pthread_kill + 10 > libsystem_kernel.dylib`__pthread_kill: > -> 0x7fff97b02f06 <+10>: jae0x7fff97b02f10; <+20> > 0x7fff97b02f08 <+12>: movq %rax, %rdi > 0x7fff97b02f0b <+15>: jmp0x7fff97afd7cd; cerror_nocancel > 0x7fff97b02f10 <+20>: retq > (lldb) bt > * thread #1: tid = 0x1c38a, 0x7fff97b02f06 > libsystem_kernel.dylib`__pthread_kill + 10, queue = 'com.apple.main-thread', > stop reason = signal SIGABRT > * frame #0: 0x7fff97b02f06 libsystem_kernel.dylib`__pthread_kill + 10 > frame #1: 0x7fff8eb604ec libsystem_pthread.dylib`pthread_kill + 90 > frame #2: 0x7fff99dee6df libsystem_c.dylib`abort + 129 > frame #3: 0x7fff8909b396 libsystem_malloc.dylib`szone_error + 626 > frame #4: 0x7fff89091373 > libsystem_malloc.dylib`small_free_list_remove_ptr + 152 > frame #5: 0x7fff8908fa7c > libsystem_malloc.dylib`szone_free_definite_size + 1790 > frame #6: 0x00010007f28c pd`freebytes(fatso=0x0001020cae00, > nbytes=2048) + 44 at m_memory.c:79 > frame #7: 0x00010001f069 pd`signal_cleanup + 105 at d_ugen.c:398 > frame #8: 0x00010001f669 pd`ugen_stop + 89 at d_ugen.c:597 > frame #9: 0x00010002d62e pd`canvas_stop_dsp + 30 at g_canvas.c:1165 > frame #10: 0x00010002cad8 pd`canvas_suspend_dsp + 40 at > g_canvas.c:1181 > frame #11: 0x00010001eaf0 pd`block_set(x=0x00010041bd40, > fcalcsize=384, foverlap=1, fupsample=1) + 48 at d_ugen.c:175 > frame #12: 0x00010007d671 pd`pd_typedmess(x=0x00010041bd40, > s=0x00010030bc60, argc=0, argv=0x7fff5fbedcf0) + 1425 at m_class.c:960 > frame #13: 0x0001000805bf pd`outlet_anything(x=0x00010041c470, > s=0x00010030bc60, argc=3, argv=0x7fff5fbedcc0) + 111 at m_obj.c:422 > frame #14: 0x0001000620bf > pd`messresponder_anything(x=0x00010041c7d0, s=0x00010030bc60, argc=3, > argv=0x7fff5fbedcc0) + 47 at g_text.c:339 > frame #15: 0x00010007d832 pd`pd_typedmess(x=0x00010041c7d0, > s=0x00010030bc60, argc=3, argv=0x7fff5fbedcc0) + 1874 at m_class.c:981 > frame #16: 0x00010007696f pd`binbuf_eval(x=0x00010041c680, > target=0x00010041c7d0, argc=1, argv=0x7fff5fbee820) + 2079 at > m_binbuf.c:773 > frame #17: 0x000100061c4e pd`message_list(x=0x00010041c7a0, > s=0x0001001221b0, argc=1, argv=0x7fff5fbee820) + 62 at g_text.c:363 > frame #18: 0x00010008221d pd`pd_list(x=0x00010041c7a0, > s=0x0001001221b0, argc=1,
Re: [PD] Crasher bug
On Mon, 2017-10-02 at 13:14 +0200, Dan Wilcox wrote: > If you want more info, you can compile pd with debugging symbols > added via: > > ./configure --enable-debug CFLAGS="-g" > > I'm going to add the "-g" back to the default debug settings as it's > needed for useful stack frame info, etc. > > If you have the Xcode command line tools installed (not the Xcode > app!), you can use lldb for similar info: http://lldb.llvm.org/lldb- > gdb.html. It can give you the exact location in the code where the > crash occurred. Thanks, Dan. Just in case this is useful to anyone, here is the backtrace on macOS: $ lldb -- pd -stderr -rt -jack -jackname tpf-client-1 -nojackconnect -inchannels 8 -outchannels 64 -open /Users/itz/pd-src/tpf/pd-tools/tpf-client.pd (lldb) target create "pd" Current executable set to 'pd' (x86_64). (lldb) settings set -- target.run-args "-stderr" "-rt" "-jack" "-jackname" "tpf-client-1" "-nojackconnect" "-inchannels" "8" "-outchannels" "64" "-open" "/Users/itz/pd-src/tpf/pd-tools/tpf-client.pd" (lldb) run Process 28469 launched: '/usr/local/bin/pd' (x86_64) Pt_Start() called slipenc: maximum packet length is 16384 iemnet - networking with Pd: [tcpclient] compiled on Dec 14 2015 at 14:29:01 copyright © 2010-2015 IOhannes m zmoelnig, IEM based on mrpeach/net, based on maxlib slipdec: maximum packet length is 16384 iemnet - networking with Pd: [udpclient] compiled on Dec 14 2015 at 14:29:01 copyright © 2010-2015 IOhannes m zmoelnig, IEM based on mrpeach/net, based on maxlib pd(28469,0x7fff79118000) malloc: *** error for object 0x1020f0c08: incorrect checksum for freed object - object was probably modified after being freed. *** set a breakpoint in malloc_error_break to debug Process 28469 stopped * thread #1: tid = 0x1c38a, 0x7fff97b02f06 libsystem_kernel.dylib`__pthread_kill + 10, queue = 'com.apple.main-thread', stop reason = signal SIGABRT frame #0: 0x7fff97b02f06 libsystem_kernel.dylib`__pthread_kill + 10 libsystem_kernel.dylib`__pthread_kill: -> 0x7fff97b02f06 <+10>: jae0x7fff97b02f10; <+20> 0x7fff97b02f08 <+12>: movq %rax, %rdi 0x7fff97b02f0b <+15>: jmp0x7fff97afd7cd; cerror_nocancel 0x7fff97b02f10 <+20>: retq (lldb) bt * thread #1: tid = 0x1c38a, 0x7fff97b02f06 libsystem_kernel.dylib`__pthread_kill + 10, queue = 'com.apple.main-thread', stop reason = signal SIGABRT * frame #0: 0x7fff97b02f06 libsystem_kernel.dylib`__pthread_kill + 10 frame #1: 0x7fff8eb604ec libsystem_pthread.dylib`pthread_kill + 90 frame #2: 0x7fff99dee6df libsystem_c.dylib`abort + 129 frame #3: 0x7fff8909b396 libsystem_malloc.dylib`szone_error + 626 frame #4: 0x7fff89091373 libsystem_malloc.dylib`small_free_list_remove_ptr + 152 frame #5: 0x7fff8908fa7c libsystem_malloc.dylib`szone_free_definite_size + 1790 frame #6: 0x00010007f28c pd`freebytes(fatso=0x0001020cae00, nbytes=2048) + 44 at m_memory.c:79 frame #7: 0x00010001f069 pd`signal_cleanup + 105 at d_ugen.c:398 frame #8: 0x00010001f669 pd`ugen_stop + 89 at d_ugen.c:597 frame #9: 0x00010002d62e pd`canvas_stop_dsp + 30 at g_canvas.c:1165 frame #10: 0x00010002cad8 pd`canvas_suspend_dsp + 40 at g_canvas.c:1181 frame #11: 0x00010001eaf0 pd`block_set(x=0x00010041bd40, fcalcsize=384, foverlap=1, fupsample=1) + 48 at d_ugen.c:175 frame #12: 0x00010007d671 pd`pd_typedmess(x=0x00010041bd40, s=0x00010030bc60, argc=0, argv=0x7fff5fbedcf0) + 1425 at m_class.c:960 frame #13: 0x0001000805bf pd`outlet_anything(x=0x00010041c470, s=0x00010030bc60, argc=3, argv=0x7fff5fbedcc0) + 111 at m_obj.c:422 frame #14: 0x0001000620bf pd`messresponder_anything(x=0x00010041c7d0, s=0x00010030bc60, argc=3, argv=0x7fff5fbedcc0) + 47 at g_text.c:339 frame #15: 0x00010007d832 pd`pd_typedmess(x=0x00010041c7d0, s=0x00010030bc60, argc=3, argv=0x7fff5fbedcc0) + 1874 at m_class.c:981 frame #16: 0x00010007696f pd`binbuf_eval(x=0x00010041c680, target=0x00010041c7d0, argc=1, argv=0x7fff5fbee820) + 2079 at m_binbuf.c:773 frame #17: 0x000100061c4e pd`message_list(x=0x00010041c7a0, s=0x0001001221b0, argc=1, argv=0x7fff5fbee820) + 62 at g_text.c:363 frame #18: 0x00010008221d pd`pd_list(x=0x00010041c7a0, s=0x0001001221b0, argc=1, argv=0x7fff5fbee820) + 77 at m_pd.c:288 frame #19: 0x00010008051f pd`outlet_list(x=0x00010041c740, s=0x0001001221b0, argc=1, argv=0x7fff5fbee820) + 111 at m_obj.c:411 frame #20: 0x0001000a143d pd`trigger_list(x=0x00010041c6d0, s=0x0001001221b0, argc=1, argv=0x7fff5fbee820) + 413 at x_connective.c:1033 frame #21: 0x00010008221d pd`pd_list(x=0x00010041c6d0, s=0x0001001221b0, argc=1, argv=0x7fff5fbee820) + 77 at m_pd.c:288 frame #22:
Re: [PD] Crasher bug
If you want more info, you can compile pd with debugging symbols added via: ./configure --enable-debug CFLAGS="-g" I'm going to add the "-g" back to the default debug settings as it's needed for useful stack frame info, etc. If you have the Xcode command line tools installed (not the Xcode app!), you can use lldb for similar info: http://lldb.llvm.org/lldb-gdb.html <http://lldb.llvm.org/lldb-gdb.html>. It can give you the exact location in the code where the crash occurred. > On Oct 2, 2017, at 12:00 PM, pd-list-requ...@lists.iem.at wrote: > > From: Roman Haefeli <reduz...@gmail.com <mailto:reduz...@gmail.com>> > To: "pd-list@lists.iem.at <mailto:pd-list@lists.iem.at>" > <pd-list@lists.iem.at <mailto:pd-list@lists.iem.at>> > Subject: Re: [PD] Crasher bug > Message-ID: <1506929734.2070.4.ca...@gmail.com > <mailto:1506929734.2070.4.ca...@gmail.com>> > Content-Type: text/plain; charset="utf-8" > > On Son, 2017-10-01 at 22:16 +0200, Roman Haefeli wrote: >> Hey all >> >> Attached patch crashes Pd, at least on Ubuntu 16.04 amd64. Pd 0.48 >> and >> 0.47-1 seems affected, but not 0.46-7. Also, I cannot make it crash >> with the current 0.48 release for Windows under Wine. I couldn't test >> it on macOS, but the original patch (not this reduced one) crashes on >> macOS 10.11 with Pd 0.48. > > I checked again and it doesn't crash on macOS, however the original > non-reduced patch does reliably crash. It seems it requires a different > reduced version for macOS. > > From 0.48, macOS 10.11: > > pd(4254,0x7fff79118000) malloc: *** error for object 0x7fc3b2096208: > incorrect checksum for freed object - object was probably modified after > being freed. > *** set a breakpoint in malloc_error_break to debug > Pd: signal 6 > gui socket 4 - closing audio... Dan Wilcox @danomatika <http://twitter.com/danomatika> danomatika.com <http://danomatika.com/> robotcowboy.com <http://robotcowboy.com/> ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Re: [PD] Crasher bug
On Son, 2017-10-01 at 22:16 +0200, Roman Haefeli wrote: > Hey all > > Attached patch crashes Pd, at least on Ubuntu 16.04 amd64. Pd 0.48 > and > 0.47-1 seems affected, but not 0.46-7. Also, I cannot make it crash > with the current 0.48 release for Windows under Wine. I couldn't test > it on macOS, but the original patch (not this reduced one) crashes on > macOS 10.11 with Pd 0.48. I checked again and it doesn't crash on macOS, however the original non-reduced patch does reliably crash. It seems it requires a different reduced version for macOS. From 0.48, macOS 10.11: pd(4254,0x7fff79118000) malloc: *** error for object 0x7fc3b2096208: incorrect checksum for freed object - object was probably modified after being freed. *** set a breakpoint in malloc_error_break to debug Pd: signal 6 gui socket 4 - closing audio... Roman signature.asc Description: This is a digitally signed message part ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Re: [PD] Crasher bug
* Roman Haefeli[2017-10-01 22:17]: > Hey all > > Attached patch crashes Pd, at least on Ubuntu 16.04 amd64. Pd 0.48 and > 0.47-1 seems affected, but not 0.46-7. Also, I cannot make it crash > with the current 0.48 release for Windows under Wine. I couldn't test > it on macOS, but the original patch (not this reduced one) crashes on > macOS 10.11 with Pd 0.48. > > The patch doesn't do anything meaningful anymore more, since I tried to > remove anything that was not needed for the crash. > > With current git master, I get the following backtrace printed to the > console, as soon as I click the bang: And I do make it crash with the same backtrace on Pd-0.48.0 from Debian's testing package on Debian amd64. ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
Trying to reproduce this in pd-l2ork. What do you mean by the last sentence? Delete all the objects? Is this on the MAIN.pd or inside text.pd? Also, unchecking the GOP option, if that applies to test, haven't we already deleted it in the previous step? Or, have you actually saved the text patch as its help file with the GOP enabled? On 09/16/2014 07:18 AM, Max wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 here is how to reproduce it: make two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd open test.pd as an abstraction. make it GOP. from MAIN rightclick and coose help the context menu to open the not existing help file. The GOP abstraction test.pd will open. Save it as test-help.pd. Delete all the objects. Uncheck the GOP option. Crash. that's in Pd 0.45.4 On 09/16/2014 08:02 PM, Max wrote: hi list, i got this at two incidents: terminate called after throwing an instance of 'boost::exception_detail::clone_implboost::exception_detail::error_info_injectorboost::lock_error ' pure virtual method called terminate called recursively but I can't reproduce it now. I was doing something like this: making two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd calling test.pd as an abstraction and going through the context menu to the not existing help file. Saving this as test-help.pd. i think that's where the crash happened. m. ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -BEGIN PGP SIGNATURE- Version: GnuPG v1 iEYEARECAAYFAlQYHIEACgkQ3EB7kzgMM6IpmQCeIBR0E2ASWDwkZ6vb7koJttSY w8cAnAx8K+XR2GXoi8GuUl1YH9TiTRB8 =H6hU -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -- Ivica Ico Bukvic, D.M.A. Associate Professor Computer Music ICAT Senior Fellow DISIS, L2Ork Virginia Tech School of Performing Arts - 0141 Blacksburg, VA 24061 (540) 231-6139 i...@vt.edu www.performingarts.vt.edu disis.music.vt.edu l2ork.music.vt.edu ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
I believe the bug, at its simplest, was this: Make an empty patch (in Pd 0.46, say), set it as GOP, save it, close it, open it again, un-set GOP, and kaboom. Amazing I never saw that before... M On Wed, Sep 17, 2014 at 05:57:13AM -0400, Ivica Ico Bukvic wrote: Trying to reproduce this in pd-l2ork. What do you mean by the last sentence? Delete all the objects? Is this on the MAIN.pd or inside text.pd? Also, unchecking the GOP option, if that applies to test, haven't we already deleted it in the previous step? Or, have you actually saved the text patch as its help file with the GOP enabled? On 09/16/2014 07:18 AM, Max wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 here is how to reproduce it: make two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd open test.pd as an abstraction. make it GOP. from MAIN rightclick and coose help the context menu to open the not existing help file. The GOP abstraction test.pd will open. Save it as test-help.pd. Delete all the objects. Uncheck the GOP option. Crash. that's in Pd 0.45.4 On 09/16/2014 08:02 PM, Max wrote: hi list, i got this at two incidents: terminate called after throwing an instance of 'boost::exception_detail::clone_implboost::exception_detail::error_info_injectorboost::lock_error ' pure virtual method called terminate called recursively but I can't reproduce it now. I was doing something like this: making two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd calling test.pd as an abstraction and going through the context menu to the not existing help file. Saving this as test-help.pd. i think that's where the crash happened. m. ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -BEGIN PGP SIGNATURE- Version: GnuPG v1 iEYEARECAAYFAlQYHIEACgkQ3EB7kzgMM6IpmQCeIBR0E2ASWDwkZ6vb7koJttSY w8cAnAx8K+XR2GXoi8GuUl1YH9TiTRB8 =H6hU -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -- Ivica Ico Bukvic, D.M.A. Associate Professor Computer Music ICAT Senior Fellow DISIS, L2Ork Virginia Tech School of Performing Arts - 0141 Blacksburg, VA 24061 (540) 231-6139 i...@vt.edu www.performingarts.vt.edu disis.music.vt.edu l2ork.music.vt.edu ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Bug confirmed with Pd-0.46.0 (test1) compiled 13:01:01 Aug 16 2014. ++ Jack Le 17/09/2014 17:10, Miller Puckette a écrit : I believe the bug, at its simplest, was this: Make an empty patch (in Pd 0.46, say), set it as GOP, save it, close it, open it again, un-set GOP, and kaboom. Amazing I never saw that before... M On Wed, Sep 17, 2014 at 05:57:13AM -0400, Ivica Ico Bukvic wrote: Trying to reproduce this in pd-l2ork. What do you mean by the last sentence? Delete all the objects? Is this on the MAIN.pd or inside text.pd? Also, unchecking the GOP option, if that applies to test, haven't we already deleted it in the previous step? Or, have you actually saved the text patch as its help file with the GOP enabled? On 09/16/2014 07:18 AM, Max wrote: here is how to reproduce it: make two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd open test.pd as an abstraction. make it GOP. from MAIN rightclick and coose help the context menu to open the not existing help file. The GOP abstraction test.pd will open. Save it as test-help.pd. Delete all the objects. Uncheck the GOP option. Crash. that's in Pd 0.45.4 On 09/16/2014 08:02 PM, Max wrote: hi list, i got this at two incidents: terminate called after throwing an instance of 'boost::exception_detail::clone_implboost::exception_detail::error_info_injectorboost::lock_error ' pure virtual method called terminate called recursively but I can't reproduce it now. I was doing something like this: making two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd calling test.pd as an abstraction and going through the context menu to the not existing help file. Saving this as test-help.pd. i think that's where the crash happened. m. ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -- Ivica Ico Bukvic, D.M.A. Associate Professor Computer Music ICAT Senior Fellow DISIS, L2Ork Virginia Tech School of Performing Arts - 0141 Blacksburg, VA 24061 (540) 231-6139 i...@vt.edu www.performingarts.vt.edu disis.music.vt.edu l2ork.music.vt.edu ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -BEGIN PGP SIGNATURE- Version: GnuPG v1 iQEcBAEBAgAGBQJUGahkAAoJEOuluecjw8GUEy0H/ROVQkbDottg7syqGGvpd8Vk FgHIWM5wu+fUf+tzFzf8t+J9fFKg/QXLmpIXtfDMEzy+1yyIsPf9Qjf2sGbpQbGH d+hDU9I5NWIaoUB56XMaFO7/sjIMf+bXNQmKNcJkR8fITX7iiWoNTB487Ej9iBhq k9PKanPy4vRj0D9gvHEu9i7j6HcqigrvefzRHb8LYLO4+Q/CW958qd1i5fQfoTO4 v+LMP2GSeO0N1g/HNYHFXlieOFf/x0MHGQ8YvQnKaHW5eXRnvdOwQZ85wZ4KacrP RbBVP+ylUHek/PM+QgBgsmfztz5WsWDrUDft4sNPmCvfbqZQjRtwuKqonErKrZM= =E8a0 -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
On 09/17/2014 11:10 AM, Miller Puckette wrote: I believe the bug, at its simplest, was this: Make an empty patch (in Pd 0.46, say), set it as GOP, save it, close it, open it again, un-set GOP, and kaboom. Amazing I never saw that before... M Thank you both for the clarification. It seems pd-l2ork is unaffected. Best, Ico On Wed, Sep 17, 2014 at 05:57:13AM -0400, Ivica Ico Bukvic wrote: Trying to reproduce this in pd-l2ork. What do you mean by the last sentence? Delete all the objects? Is this on the MAIN.pd or inside text.pd? Also, unchecking the GOP option, if that applies to test, haven't we already deleted it in the previous step? Or, have you actually saved the text patch as its help file with the GOP enabled? On 09/16/2014 07:18 AM, Max wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 here is how to reproduce it: make two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd open test.pd as an abstraction. make it GOP. from MAIN rightclick and coose help the context menu to open the not existing help file. The GOP abstraction test.pd will open. Save it as test-help.pd. Delete all the objects. Uncheck the GOP option. Crash. that's in Pd 0.45.4 On 09/16/2014 08:02 PM, Max wrote: hi list, i got this at two incidents: terminate called after throwing an instance of 'boost::exception_detail::clone_implboost::exception_detail::error_info_injectorboost::lock_error ' pure virtual method called terminate called recursively but I can't reproduce it now. I was doing something like this: making two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd calling test.pd as an abstraction and going through the context menu to the not existing help file. Saving this as test-help.pd. i think that's where the crash happened. m. ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -BEGIN PGP SIGNATURE- Version: GnuPG v1 iEYEARECAAYFAlQYHIEACgkQ3EB7kzgMM6IpmQCeIBR0E2ASWDwkZ6vb7koJttSY w8cAnAx8K+XR2GXoi8GuUl1YH9TiTRB8 =H6hU -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -- Ivica Ico Bukvic, D.M.A. Associate Professor Computer Music ICAT Senior Fellow DISIS, L2Ork Virginia Tech School of Performing Arts - 0141 Blacksburg, VA 24061 (540) 231-6139 i...@vt.edu www.performingarts.vt.edu disis.music.vt.edu l2ork.music.vt.edu ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -- Ivica Ico Bukvic, D.M.A. Associate Professor Computer Music ICAT Senior Fellow DISIS, L2Ork Virginia Tech School of Performing Arts - 0141 Blacksburg, VA 24061 (540) 231-6139 i...@vt.edu www.performingarts.vt.edu disis.music.vt.edu l2ork.music.vt.edu ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 here is how to reproduce it: make two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd open test.pd as an abstraction. make it GOP. from MAIN rightclick and coose help the context menu to open the not existing help file. The GOP abstraction test.pd will open. Save it as test-help.pd. Delete all the objects. Uncheck the GOP option. Crash. that's in Pd 0.45.4 On 09/16/2014 08:02 PM, Max wrote: hi list, i got this at two incidents: terminate called after throwing an instance of 'boost::exception_detail::clone_implboost::exception_detail::error_info_injectorboost::lock_error ' pure virtual method called terminate called recursively but I can't reproduce it now. I was doing something like this: making two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd calling test.pd as an abstraction and going through the context menu to the not existing help file. Saving this as test-help.pd. i think that's where the crash happened. m. ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -BEGIN PGP SIGNATURE- Version: GnuPG v1 iEYEARECAAYFAlQYHIEACgkQ3EB7kzgMM6IpmQCeIBR0E2ASWDwkZ6vb7koJttSY w8cAnAx8K+XR2GXoi8GuUl1YH9TiTRB8 =H6hU -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 09/16/2014 10:26 PM, IOhannes m zmölnig wrote: On 09/16/2014 01:18 PM, Max wrote: here is how to reproduce it: make two pd documents, let's say MAIN.pd and text.pd. then, in MAIN.pd open test.pd as an abstraction. make it GOP. from MAIN rightclick and coose help the context menu to open the not existing help file. The GOP abstraction test.pd will open. Save it as test-help.pd. Delete all the objects. Uncheck the GOP option. Crash. that's in Pd 0.45.4 On 09/16/2014 08:02 PM, Max wrote: hi list, i got this at two incidents: terminate called after throwing an instance of 'boost::exception_detail::clone_implboost::exception_detail::error_info_injectorboost::lock_error your steps to reproduce and your backtrace are unrelated. boost is a C++ helper library, while Pd is written in pure C. so the error is either related to a C++-external you are loading (not Gem, as it does not use boost either), or it indicates memory corruption. okay, but can you reproduce it? -BEGIN PGP SIGNATURE- Version: GnuPG v1 iEYEARECAAYFAlQYWWoACgkQ3EB7kzgMM6KfpgCfcWZuIps6TMfnRXa4oO0KrgMi B6QAn3kMPs+6gO61ePID0qig/YGx+5mE =g5MX -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 09/17/2014 01:09 AM, Miller Puckette wrote: okay, but can you reproduce it? I can :) - I hadn't read far enough down the original mail to see that this is indeed a Pd vanilla bug - no externs at all. This might be very helpful as I've heard varios things about GOP trouble I can't get to the bottom of. I'll give it a look next time I can regain sanity for a few minutes. Ok, great. Let me turn this into a feature-request: If I have an abstraction without help file, the context menu entry should read: Create help, instead of Help. If I choose this a new blank document named {abstraction_name}-help.pd should be opened with the abstraction used inside. Even better would be if the unofficial help-template was used. (!) with the awesomeness of [pd meta] in it for the search plugin. 1.5 ct. m. -BEGIN PGP SIGNATURE- Version: GnuPG v1 iEYEARECAAYFAlQYaNIACgkQ3EB7kzgMM6L7lwCcCJZQNB5AIK7fPiF9h6WW6VNz f7QAnA7Un+PwVzZquDX6aHaE6d9fvY78 =rXT1 -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] crasher bug
OK -- I fixed the crash bug and also suppressed opening the patch as its own help file (that was unintentional anyway). OTOH I didn't add a pre-check to see if the help file was available when the object gets right-clicked on... that's a bigger change and on some systems it might make the menu slow to pop up. cheers M On Wed, Sep 17, 2014 at 01:44:02AM +0900, Max wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 09/17/2014 01:09 AM, Miller Puckette wrote: okay, but can you reproduce it? I can :) - I hadn't read far enough down the original mail to see that this is indeed a Pd vanilla bug - no externs at all. This might be very helpful as I've heard varios things about GOP trouble I can't get to the bottom of. I'll give it a look next time I can regain sanity for a few minutes. Ok, great. Let me turn this into a feature-request: If I have an abstraction without help file, the context menu entry should read: Create help, instead of Help. If I choose this a new blank document named {abstraction_name}-help.pd should be opened with the abstraction used inside. Even better would be if the unofficial help-template was used. (!) with the awesomeness of [pd meta] in it for the search plugin. 1.5 ct. m. -BEGIN PGP SIGNATURE- Version: GnuPG v1 iEYEARECAAYFAlQYaNIACgkQ3EB7kzgMM6L7lwCcCJZQNB5AIK7fPiF9h6WW6VNz f7QAnA7Un+PwVzZquDX6aHaE6d9fvY78 =rXT1 -END PGP SIGNATURE- ___ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list