Set Crashes
Brought to you by:
set
From: J.B. L. <let...@un...> - 2004-12-09 09:22:32
|
Dear Set: This is very complicated now. 1. First I re-configued and re-compiled and re-installed tvision and setedit as per your instructions with the CFLAG and CXXFLAG environment variables set. I used a plain configure, no options. With this compilation I could NOT crash setedit, with or without gdb. Neither the editor.exe, nor, later the setedit binary once I had installed it. I tried compiling again with the no-intl option for both tvision and setedit. No problems, no crashes. It is definitely the new binary running, and the "help/About" gives the 5.4 identification. So: I tried WITHOUT the CFLAG variables set (configure, make make install) both the editor and the tv libraries, and could not crash setedit either with the editor.exe or the re-re-installed setedit. Please note that nothing has changed on the computer in the meantime (at least as far as I can recall, and I have wrkaced my brains and my log files); I haven't altered things for months and months now. BUT: I had accidentally deleted the source tarball for tvision last week, and had to download another to make these tests: though whether there have been any changes I wouldn't know. OK: finally, I re-compiled both tvision and setedit from the beginning, without the CFLAG variables set, but WITH "--no-intl" as a configure option for both tvision and setedit. This time it crashed quite happily and consistently. The --no-intl is my normal option. I hoe that is a clear account of what I have done. If something seems illogical or inexplicable, please don't hesitate to ask me to repeat something. Below are the results from these crashes. Note: I normally work from the console. There were problems with recording the crashes there however (see below), so I switched to Xwindows (where you have a marvellous interface, may I say), and recorded with no difficulty the FIRST report below. However, under Xwindows, partly becasue of the corruption of the keyboard, I couldn't reproduce all the styles of crash: so I returned to the console with the "hack" which I explain next to get the other data as best I could. So: the second group are a bit chaotic perhaps, since, once gdb runs editor.exe and it crashes it grabs the mouse and the alt and ctrl keys. So that I had to keep quitting gdb after each crash, paste from a garbled screen, fix the garbage, re-run gdb editor.exe, run "continue" past the crashes I already had, and quit and paste the next step, again from the garbled screen and so on. I have tried to be careful. I have deleted repeated material. Finally: the crashes always occur when I try to change console (alt arrow, or alt F1 etc); AND/OR when I fill in a box, such as the TAGS dialogue which pops up on Ctrl-Tab: space to check the "don't show next time" box; return to get rid of the dialogue -- crash. Ditto with "search": enter the expression, hit return, crash. Also with the dialogue mentioned below Editor General:check box, press enter, crash. Sorry to bang on so long: I thought I'd try and give all the information I could think of. Let me know if I can or should do anything else. 1. [Under Xwindows] Program received signal SIGSEGV, Segmentation fault. 0x40073b67 in TStringCollectionCIntl::freeItem () from /usr/lib/librhtv.so.2.0.3 (gdb) continue Continuing. I'm sorry but I did something wrong :-( Look in the file /home/nemle01/.setedit/erY77Yzd for unsaved buffers and information. Please send this file to the author (se...@ie...). Program received signal SIGSEGV, Segmentation fault. 0x080b262f in StackDBG_GCCDumpStack () 2. [Under Console] [this next as a consequence of pressing Alt+left arrow, to try and change consoles] Program received signal SIGUSR1, User defined signal 1. 0x401fd16e in __select () at__select:-1 -1 __select: No such file or directory. [Ditto, but this is the next "continue" Alt+right this time] Program received signal SIGUSR2, User defined signal 2. 0x401e3431 in __libc_nanosleep () at __libc_nanosleep:-1 -1 __libc_nanosleep: No such file or directory. in __libc_nanosleep (gdb) q Program received signal SIGUSR1, User defined signal 1. 0x401fd16e in __select () at __select:-1 -1db) co__select: No such file or directory. Continuiin __select [trying to alter unix backups in tool/options/editor general:] Program received signal SIGUSR1, User defined signal 1. 0x40184226 in sigaction () from /lib/libc.so.6 (gdb) continue continuing [This message repeated but with "SIGUSR2 User defined signal 2."] Hope this helps, All the best, Julian Dr J.B. Lethbridge English Seminar University of Tuebingen Wilhelmstrasse 50 Tuebingen 72074 Germany |