Segmentation fault adding a new layer with php-mapscript (4.6beta3)

[Bill Binko]

On Mon, 6 Jun 2005, Nicol Hermann wrote:

> Bill,
> 
> Am Montag, den 06.06.2005, 10:19 -0400 schrieb Bill Binko:
> > On Mon, 6 Jun 2005, Nicol Hermann wrote:
> > 
> 
> <cut>
> 
> > 
> > > Sadly the segfault persists.
> > 
> > bummer
> > 
> > > 
> > > Maybe of note PHP is compiled as cgi with the flag
> > > '--with-regex=system'.
> > 
> > That was what the configure script detected, so that's good.
> > 
> > At this point, it might not be the regex problem.. could you run it in gdb 
> > and post the results?  That would show us what the corruption was.

Hmm..

> 0x4114ea30 in layerObj_new (map=0x41f88008) at mapscript_i.c:410
> 410         map->layerorder[map->numlayers] = map->numlayers;
> (gdb)

Looks like the layerorder array has not been malloc'd or that the sizeof 
the struct is munched (so that map->layerorder doesn't point to the right 
place).

The first would be a problem with initMap not being called or someone 
overwriting layerorder.  The second could be the mismatch in regex.  
However, that's looking less likely now.

Could you try building with a simpler configure line?  I do not have all 
of the components (Oracle etc) that are needed for yours.

Here is what I just tested with (and it worked for me):

./configure --with-xpm=/usr/X11R6 --with-proj --with-postgis 
     --with-gd=/gis/libs/  --prefix=/usr --with-curl --with-ogr=/usr 
     --with-wmsclient --enable-debug 
     --with-http=/usr/sbin/httpd2 --with-php=/usr --with-wmsserver

Obviously, you'll need to change the --with-xpm etc.

Try that and see if it still crashes: if not, we'll find the offensive 
option.  If so, at least we'll be on the same page.

Bill