Denis Mysenko wrote: > > Doesn't compile even on last FreeBSD CVS. I mean that you should take aspseek from CVS, not FreeBSD. I have fixed all the problems that were in 1.2.6 - absence of sys/user.h, MAP_ANONYMOUS, etc. Can you get the latest CVS (see http://www.aspseek.org/cvs.html) and see if it works for you?
> Still need to replace "#include <sys/user.h>" in buddy.h with > "#include <sys/param.h>" line and MAP_ANONYMOUS with MAP_ANON in > buddy.cpp (twice) to get it compiled. > > Replacing two lines you specified helps - index doesn't coredump so fast, > but cryes from time to time: > index in free(): warning: junk pointer, too high to make sense. > And after several minutes we get: > Program received signal SIGSEGV, Segmentation fault. > void __introsort_loop<CULONG *, CULONG, int> (__first=0xbfaf7870, __last=0xbfaf78c8, >__depth_limit=9) at > /usr/include/g++/stl_algo.h:1283 > 1283 _RandomAccessIter __cut = > > index -D gives following output: > Program received signal SIGSEGV, Segmentation fault. > CDeltaFiles::Save (this=0xbfbfe910, nfile=4, readTime=@0xbfbfe724, >sortTime=@0xbfbfe72c, memory=@0xbfbfe738) at deltas.cpp:886 > 886 pbuf += (pbuf[2] & 0x3FFF) + 3; > > searchd still cores if there are "-" or few languages in query. Seems that this still needs investigation. The problem is I don't use FreeBSD and don't have a spare box to debug on. But I still want ASPSeek to be usable on FreeBSD and don't want to be ignorant to FreeBSD-related problems, so if you (or anybody else) will help, it would be great. -- [EMAIL PROTECTED] ICQ 7551596 Phone +7 903 6722750 Hard work may not kill you, but why take chances? --
