[Reposted with updates for 3.0b4-dev behavior]

I've been experimenting with odbc_prepare() and odbc_execute()
to try to figure out how to use them, and have succeeded only
in generating application errors when calling odbc_execute().
I don't know for sure that I'm using them right, but they
should at least complain and exit gracefully, not crash in
such an ugly way.

Here's my code:

  $result = odbc_prepare($conn,"{call sp_asp_dmsGetArtist (?)}");
  $parm[0] = 303;
  $result = odbc_execute($result, $parm);

If I comment out odbc_execute(), the code completes nicely,
but leaving it in causes a crash.  According to SQLTrace, 
the query *is* getting sent to the server, which implies 
that the crash occurs after the SQLExecute() at line 626 
of unified_odbc.c (3.0b3-dev).

If I change (?) -> (?, ?) in the first line and insert
  $parm[1] = 0;
after the second line, I get the same results.

However, if I change (?) -> (?, ?) and leave $parm[1] unset,
PHP still crashes, but the query is never sent to the server.
In this case, PHP should catch the mismatch in parameters 
at line 599 and return immediately, which explains why
SQLExecute() isn't called.  Unfortunately, that doesn't 
explain why PHP still crashes.

**** Update for 3.0b4-dev ****

Using
  $result = odbc_prepare($conn,"{call sp_asp_dmsGetArtist (?, ?)}");
while leaving out
  $parm[1] = 0;
no longer causes a crash.  It now properly returns with:
  Warning:  Not enough parameters (1 should be 2) given

In the cases that still cause a crash, the last output from
PHP is:
  <br>
  <b>Warning:</b>
which implies that PHP is dieing inside of php3_error()?!