Chris Humphries wrote:
> In the first test, the program crashed whilst indexing the first HTML (the
> PHRASEm build managed to build an index but I was unable to search that
> index - PHRASEm did not crash). In the second test the index built ok but
> the search hung.
This could be fun. Did you modify the config file between the two
instances? I am still seeing Index header problems on BSD. It very
well could still be an issue on Windows, as well.
> This is very interesting, because the reason I am trying to use the new
> Swish is that swish_132.exe is currently crashing under NT. The crash I am
Really. This I have not seen. I used 1.3.2 almost exclusively on
Windows NT. Seems like this could be an annyoying problem to find. Are
you getting the same problems on 1.3.2 as 2.0b2? (I'm not sure which
would be hardest to debug :-)
> getting is an illegal write through esi just into a function. Without
> debugging information I could not say what that function is. I can't quite
> read the machine code because I am not used to reading pipelined code, but
> it appears to be the 2nd or 3rd variable updated - probably a loop counter
There are loads of loops in there. Almost any problem is going to be in
some sort of loop.
> Could you send me an .exe with debugging symbols to see if I can trace this
> crash? I realise that the debug .exe will have a different layout and that
I certianly can get you a debug build. I normal do both debug and
release just in case. For some stupid reason I deleted the debug
build. I should be able to build a debug executable in about 12 hours.
>From now on I'll make sure that debug and release are both available at
the same time.
I was seeing differences between debug (ran) and release (crashed) on
Win 98 in some very early builds. That problem disappeared as the code
stabilized a bit.
Dave's Web - http://www.webaugur.com/dave/
Dave's Weather - http://www.webaugur.com/dave/wx
ICQ Universal Internet Number - 412039
E-Mail - firstname.lastname@example.org
Received on Mon Jul 3 04:26:28 2000