Re: make test result (4.1.0RC CVS)
| From: | Yasuo Ohgaki | Date: | Mon, 12 Nov 2001 02:36:00 +0000 |
| Subject: | Re: make test result (4.1.0RC CVS) | ||
| References: | 1 2 | Groups: | php.qa |
| Request: | Send a blank email to php-qa+get-3808@lists.php.net to get a copy of this message | ||
Sander Roobol wrote:
For #11389 array_walk does not take into account if target array has been changed or not. I'm not sure if it is feasible to detect that target_hash is deleted/changed in user function or not. When php_array_walk() get pointer to hash table, it contains valid pointer to hash, but it will be invalid if user does like the script in the bug report. (assign null to target array(hash), for example) The script didn't crash on my system. PHP exits from script silently. However, it's a cause of memory problem and PHP is not supposed to behave like this... For the time being, documentation is better to note about that users are not supposed to change target array in user defined function for array_walk/array_filter. (I suppose array_filter has the same issue) -- Yasuo OhgakiFollowing critical(?) problems are seems to be fixed in latest 4.1.0RC CVS. - zlib output compression crash - segfault at the shutdown when --enable-debug is enabled (mostly fixed?) Rather critical bugs, that I know & verified exists in both 4.1.0 and 4.2.0, are - #8441 (register_globals=on and track_vars=on produce strange results for arrays) - #13703 (PHP allows function redefinition)What about #11389 and #12776? Both are bugs about crashes using array_walk. Don't know if they're related, but it seems someone needs to take a look at the array_walk function...