Re: [PHP4BETA] cvs: php4 /ext/odbc/ php_odbc.c

Nick Gorham wrote:
> 
> Hmm
> 
> not sure where the "while ( henv != SQL_NULL_HENV )" has come from there, I added
> the code to retrieve all the errors, And I think it should be allowed to stay. You
> often get a "driver not capable" error when connecting, this hides the real errors
> that may happen. When I added this the code read
> 
> #if !defined (HAVE_ADABAS)
>     do {
> #endif
>         ret = SQLError(henv, conn, stmt, state,
>                 &error, errormsg, sizeof(errormsg)-1, &errormsgsize);
>         if (func) {
>             php_error(E_WARNING, "SQL error: %s, SQL state %s in %s",
>                    errormsg, state, func);
>         } else {
>             php_error(E_WARNING, "SQL error: %s, SQL state %s",
>                     errormsg, state);
>         }
> #if !defined (HAVE_ADABAS)
>     } while (SQL_SUCCEEDED(ret));
> #endif
> 
> The !defined HAVE_ADABAS was added later. Would anyone object if it went back to
> that state ?
> 

I see your point, but many drivers are buggy regarding SQLError(), since
the do ... while above causes them to go into an infinte loop, never
returning SQL_NO_DATA_FOUND. This behaviour comes with Adabas (therefore
HAVE_ADABAS which has magically dissapeared and was replaced by "while (
henv != SQL_NULL_HENV )" by whoever), DBMaker, and probably Oracle
(Win32).
I think this "feature" sent IIS on my NT box to Nirvana while testing
the 4.0 release binaries ;-).

-Andreas