Hi David
I remember, I hit it myself the FSStat structure problem and fixed it
delegating
the business of getting file info to APR only.
No, it doesn't seem the problem can be treated in the way we fixed the
stat call.
It's not something you can track down by putting a trace on a memory
location
to snare the problem, waiting for the offending code to tamper with that
location.
In this case the arguments before the call are OK and as soon as you
press the
'step into' key and control passes to ap_get_module_config the argument
values
have changed....
Speaking with a colleague during lunch he came up with a suggestion. Since
the Tcl_PkgRequire call for RivetLib seems to introduce the bug this
must have
to do with names and symbols run time resolution. I'm trying to single
out possible
names that could be clashing, but without success so far.
thanks anyway
-- Massimo
On 02/08/2011 01:03 PM, David Welton wrote:
I once had a Rivet problem that took me several days to track down: in
the end it turned out that Tcl and Apache were linking to different
size structs for something (stat or fstat or something like that?),
and sooner or later, something tried to write N bytes (as per its idea
of that struct) into M bytes (the other system's definition of the
struct), which wrote over the end of it and onto something else, that
then crashed later.
I have no idea if it's related to what you're seeing, just thought I'd
throw it out there for consideration.
---------------------------------------------------------------------
To unsubscribe, e-mail: rivet-dev-unsubscr...@tcl.apache.org
For additional commands, e-mail: rivet-dev-h...@tcl.apache.org
