PHP IDE :: NuSphere Forums :: View topic - listener.startsession(...) returns hresult = DBGSE_WRONGREQ

Posted: Tue Oct 26, 2004 2:02 pm

Hi Folks,

I built dbg-cli and I'm steping through with gdb. I'm running on the same box as php. I've entered the following to dbg-cli:

Quote:

set mode on
set port 7869
set mapurlroot http://localhost
set maplocalroot /var/www/html
set mapremoteroot /var/www/html
file /var/www/html/index.php
run

I set a break at do_run(); command.cpp:629 which is the call to listener.startsession(...) but this call returns hresult = DBGSE_WRONGREQ.

I haven't been able to determine what he wants to be happy.

Any thoughts?

Thanks,

Chris.

Posted: Tue Oct 26, 2004 7:02 pm

Hi Folks,

The failure seems to be in open_dbg_session() and it appears to have something to do with parsing a regular expression, but it is not clear to me what is wrong.

I have found other posts that indicated that both dbg and dbg-cli need to use the same dbg_net.[ch] and I have done that, but I haven't found anything discussing this parsing problem. Anybody have any ideas?

Thanks,

Posted: Tue Oct 26, 2004 7:22 pm

Hi Folks,

More specifically, it is an open_http_session ...

Posted: Tue Oct 26, 2004 7:55 pm

Hi Folks,

Regcomp is returning 0 while trying to compile the regular expression during the first invocation. I have NO idea what is wrong with it and I don't seem to be able to step through it, since I apparently don't have my libc compiled with symbols. Confused

More to the point, I don't know how to get around this...

Posted: Wed Oct 27, 2004 3:53 pm

did you compile with regexp library that comes with dbg-cli or with system supplied ? Please check gcc logs.

Posted: Wed Oct 27, 2004 4:23 pm

I ran "make clean; make > make.log 2>&1" and I investigated.

I see explicit -I../pcre in the compilation lines and I see -L/usr/local/src/dbg-cli-2.11.0/pcre. So I conclude that I am using the supplied code and not the system default code.

Does this answer your question sufficiently?

I also believe that the error is in regexec, not regcomp, but I am only confident and not certain. I will need to refresh a lot of old knowledge to get a more precise answer, but I am eager to do that.

Posted: Wed Oct 27, 2004 5:03 pm

right.
well let's trace what's there.
open UrlParse.cpp and replace the following code

Code:

if (ferr)
return;

with

Code:

if (ferr) {
DBGTRACE(("regexec for \"%s\" failed with %d\n", url.c_str(), ferr));
return;
}

and add the following lines onto the top of this file

Code:

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include "dbg_net.h"
#include "dbgbase_globals.h"
#include "dbgbase_classes.h"

Posted: Wed Oct 27, 2004 6:32 pm

I assumed that in addition to the changes, I needed to define _TRACE, which I did but this exposes a problem. I get an error on line 51 of dbgbase_globales.cpp:

Quote:

char buf[1024];
vsnprintf(buf, sizeof(buf), str, args);
cerr << buf; <======================
if (tracelogger!=NULL)
tracelogger(buf);

Quote:

dbgbase_globals.cpp:51: `cerr' undeclared (first use this function)

Any help here?

Posted: Wed Oct 27, 2004 9:33 pm

I added #include <iostream> and changed cerr to std::cerr and I was able to complete make. I ran with the following commands:

Quote:

set mode on
set port 7869
set mapurlroot http://localhost
set maplocalroot /var/www/html
set mapremoteroot /var/www/html
file /var/www/html/index.php

and I got the following results:

Quote:

dbg>run
ThreadBase::ThreadBase
regexec for "http://localhost/index.php" failed with 1
Segmentation fault

Does this provide any elightnement?

Posted: Thu Oct 28, 2004 11:15 am

looks like gcc 3.x "feature"...
with 2.96 this code worked for years.

Posted: Thu Oct 28, 2004 12:15 pm

Can you offer any more? For example, what is happening and how do I prevent it or work around it?

Posted: Thu Oct 28, 2004 2:30 pm

one guy posted there that he has successfully compiled with system supplied pcre. Please walk through the forums for the post.