tcl-mod_dtcl mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Craig Huckabee" <>
Subject Re: mod_dtcl on Windows98 again
Date Wed, 26 Dec 2001 18:16:59 GMT

> Wojciech Kocjan <> writes:
> > > Tried debugging it?
> > How? ;-)
> > I never debugged a thing on Windows...
> Neither have I - maybe someone else can comment?  Does it crash, by
> the way, when you run it with '-X' or whatever the windows equivalent
> is?

-X is the same on Win32.

Debugging is a problem.  Since Apache is built with VC++ and mod_dtcl is
built with MingW32, and they don't share a compatible debug symbol format,
the debugger gets "lost" when trying to trace program execution.  So, it is
impossible to trace a problem in the traditional sense using a debugger.

I got some good results by using GDB to attach, run, then print out the
actual exception with breakpoint, then printing out the call stack.  That
gave me a good ball park as to where the problems were.  So far, though,
every problem I've hit has been related to a bad script, configuration file,
some other module, or the machine's setup.

Those playing along at home might ask "Why not build mod_dtcl with VC++ ?"
It can be built that way, but the binary produced is unusable.  I tried
tracking that one down once, ended up getting nowhere.  I can say it isn't
in the optimizer - a bloated unoptimized binary produced by VC++ doesn't
work either.  Seems like there was some pointer mangling going on in the
compiler, but it was over a year ago when I was actively looking at it.

Since Tcl & mod_dtcl can be built with MingW32, I think it'd be better to
have the Apache sources patched to support that build environment.  I
started coming up with all of the patches once, but stopped due to lack of
time on my behalf and apparent lack of interest from the Apache core (no
e-mail response).

What version of Apache are you using, btw ?


View raw message