I want to generate an initial aff file using affixcompress and then later be able to use much to generate new dictinary files if words are added to the word list.
But, currently munch segfaults even when using the initial words file and generated aff file.
LC_ALL=hi_IN.UTF-8 sort fixed-words.txt | LC_ALL=hi_IN.UTF-8 uniq > hin
LC_ALL=hi_IN.UTF-8 affixcompress hin 500
LC_ALL=hi_IN.UTF-8 munch hin hin.aff
munch segfaults even in the above case where the aff file has been generated by affixcompress without error.
parsing line: SFX 497 Y 1
parsing 4 entries 1
affix: नशील 12, strip: 0
stable 460 num is 1
parsing line: SFX 408
Program received signal SIGSEGV, Segmentation fault.
0x76dab551 in strcat () from C:\WINDOWS\SysWOW64\msvcrt.dll
(gdb) backtrace
(gdb) Cannot access memory at address 0xffff9ca8
(I work on MSYS2 so I know some more details about the build system / environment - Pacman/MSYS2)
Any chance you can build it with options of 'debug' and '!strip', that way the callstack will hopefully be more useful.
Hi Ray,
I built it again with 'debug' and '!strip'.
now I get..
stable 460 num is 1
parsing line: SFX 408
Program received signal SIGSEGV, Segmentation fault.
0x76dab551 in strcat () from C:\WINDOWS\SysWOW64\msvcrt.dll
(gdb) backtrace
0 0x76dab551 in strcat () from C:\WINDOWS\SysWOW64\msvcrt.dll
1 0x0028fb58 in ?? ()
(gdb) Cannot access memory at address 0xffff9ca8
Looks like this is same as before. Please suggest if I can try some other commands to get more info.
Last edit: shreeshrii 2014-10-10
The PKGBUILD needs to be fixed so that it supports debug properly, look at some of the other PKGBUILD files to get an idea what needs to be done (search for check_option "debug").
==> Starting build()...
configure: WARNING: unrecognized options: --enable-debug
We should take this off here and add as an issue as we're just adding noise to hunspell's bugtracker, if you want to, please create an issue on https://github.com/Alexpux/MINGW-packages/issues
https://github.com/Alexpux/MINGW-packages/issues/247
Found it!:
.. there's no checking or asserting when these limits are exceeded.
Please find attached two patches for these issues.
Last edit: Ray Donnelly 2014-10-10
http://openoffice.2283327.n4.nabble.com/How-to-get-list-of-valid-word-in-hunspell-td2791763.html
"Munch and unmunch work only for myspell dictionaries without any
hunspell-specific additions."
Maybe that explains why it does not work with .aff files generated by affixcompress.