Re: lynx-dev bug in dev25

[Leonid Pauzner]

>> The code in UCGetLYhndl_byMIME() returns a -1 if the character set in
>> question is not found.  Other code uses the return code value without
>> ever looking for a bad value.  This causes a core dump in
>>   [1] strncmp(0xffdbfda0, 0x240260, 0x10, 0x0, 0x49, 0x57657374), at 
>> 0xef5e40b0
>> =>[2] HTMLSetRawModeDefault(i = -1), line 601 in "LYCharSets.c"
>>   [3] character_set_fun(value = 0xefffdf55 "ISO Latin 1"), line 532 in "LYRea
> dCFG.c"
>>   [4] read_cfg(cfg_filename = 0x2d5e18 "/projects/intranet/lib/lynx//lynx.cfg
> ", parent_filename = 0x2a426c "main program", nesting_level = 1), line 1095 in
> "LYReadCFG.c"
>>   [5] main(argc = 1, argv = 0xefffe734), line 1243 in "LYMain.c"
>> (
> thanks (this doesn't sound like a new bug)
Yes, we just got an opportunity to see the bug :-)

It is easy to solve:
we have few entry points for possible error (typo) -
remote document (http/meta),
display_charset(lynx.cfg/userdefs.h/.lynxrc/command line)
assume_charset (lynx.cfg/command line) + other assumptions.

All of them should use UCGetLYhndl_byMIME().

>> Larry W. Virden                 INET: address@hidden