Re: Re: Bug #17892: Error Handling for eval()

Markus, The problem is that compile_string() will add functions and classes to PHP's global structures. Also, if there's a parse error it can leave the internal compiler structures in an unstable state. I can't think of any quick and easy way of support this kind of lint feature nor eval() recovery. Andi At 11:59 AM 6/21/2002 +0200, Markus Fischer wrote:

    Hi,

    I think you're idea is pretty good, however not the way you
    engaged the current behaviour imho.

    PHP also supports 'lint'ing of source code aka. checking for
    parserer errors. It's just a few minutes work to provide the
    same functionality for a new function:

        $retval = lint($code);

    This would also throw parserer errors right at you, if you
    don't want, use '@' to supress it. It could be named
    'eval_test' as someone on IRC (hey guys, btw:) suggested, or
    an optional argument to eval() to do only linting of the
    source, whatever.

    I've attached a small patch against
    Zend/zend_builtin_functions.c from CVS a few days ago (but it
    can be done everywhere, e.g. ext/standard ). I didn't
    commited it because a) I wasn't unsure if ppl want it and b)
    if it's the proper way to do it that way ("it works for me").

    Any of the developers care reviewing, (fixing,) commiting ?
    :)

    - Markus

On Fri, Jun 21, 2002 at 08:57:43AM -0000, Sky@quit-clan.de wrote :

From:             Sky@quit-clan.de

Operating system: Any

PHP version:      4.2.1
PHP Bug Type:     Feature/Change Request

Bug description: Error Handling for eval() I want to handle Parse_Errors and similar Fatals with my custom error handler when they occur in eval()'d code. This is necessary when i want to execute custom code in my console script and prevent this code from crashing my entire program if there is only a simple typo in there... I have achieved this by a little hack in the "zend_execute.c" but i don't know if this handles any errors right that don't occur in eval'd() code... This is what i changed (to even prevent a segmentation fault) (zend_execute.c, line 1563): if (zend_hash_find(active_function_table, function_name->value.str.val, function_name->value.str.len+1, (void **) &function)==FAILURE) { zend_error(E_USER_ERROR, "Call to undefined function: %s()", function_name->value.str.val); return; } -- Edit bug report at http://bugs.php.net/?id=17892&edit=1 --

Fixed in CVS:        http://bugs.php.net/fix.php?id=17892&r=fixedcvs
Fixed in release:    http://bugs.php.net/fix.php?id=17892&r=alreadyfixed
Need backtrace:      http://bugs.php.net/fix.php?id=17892&r=needtrace

Try newer version: http://bugs.php.net/fix.php?id=17892&r=oldversion Not developer issue: http://bugs.php.net/fix.php?id=17892&r=support Expected behavior: http://bugs.php.net/fix.php?id=17892&r=notwrong

Not enough info:     http://bugs.php.net/fix.php?id=17892&r=notenoughinfo
Submitted twice:     http://bugs.php.net/fix.php?id=17892&r=submittedtwice
register_globals:    http://bugs.php.net/fix.php?id=17892&r=globals

-- GnuPG Key: http://guru.josefine.at/~mfischer/C2272BD0.asc

Did I help you?    http://guru.josefine.at/wish_en

Konnte ich helfen? http://guru.josefine.at/wish_de "uhmm.. the dates in the bug db.. aren't they printed a bit wrong, i mean, did i miss when we changed to 53 days/month ( +2002-02-53) ? =P - N0v3ll -- PHP Development Mailing List <http://www.php.net/> To unsubscribe, visit: http://www.php.net/unsub.php