# PHP user session handlers - Thread: https://externals.io/message/26242 - Original archive: https://news-web.php.net/php.internals/26242 - Started by Glenn Richmond on 2006-10-25T09:04:23+00:00 - Messages: 9 - Reply by email to internals@lists.php.net - More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt --- ## #26242 — Glenn Richmond — 2006-10-25T09:04:23+00:00 URL: https://externals.io/message/26242#26242 Hi guys, Just wondering what the status of the user-defined session handlers is? I've tested with 5.2rc4 and there is an issue that the references to the user functions get erased when the user calls session_write_close or session_destroy. This means that the system reverts to an alternative handler. Any idea if this issue has been resolved or if not, how it could be resolved? It basically makes large-scale user-defined session handlers impossible (for general website handling). It has been discussed previously, but no solution has been proposed as of yet. Glenn. --- ## #26243 — Antony Dovgal — 2006-10-25T09:10:05+00:00 URL: https://externals.io/message/26242#26243 · In reply to: #26242 > Hi guys, > > Just wondering what the status of the user-defined session handlers is? I don't think it makes any sense to ask such questions on the list, for the answer in most cases is "they are doing fine, thank you". Just grab the latest snapshot, test it and report any issues you see. > I've tested with 5.2rc4 and there is an issue that the references to the > user functions get erased when the user calls session_write_close or > session_destroy. This means that the system reverts to an alternative > handler. > > Any idea if this issue has been resolved or if not, how it could be > resolved? It basically makes large-scale user-defined session handlers > impossible (for general website handling). It has been discussed > previously, but no solution has been proposed as of yet. -- Wbr, Antony Dovgal --- ## #26244 — Lukas Kahwe Smith — 2006-10-25T09:10:17+00:00 URL: https://externals.io/message/26242#26244 · In reply to: #26242 Glenn Richmond wrote: > Hi guys, > > Just wondering what the status of the user-defined session handlers is? > > I've tested with 5.2rc4 and there is an issue that the references to the > user functions get erased when the user calls session_write_close or > session_destroy. This means that the system reverts to an alternative > handler. > > Any idea if this issue has been resolved or if not, how it could be > resolved? It basically makes large-scale user-defined session handlers > impossible (for general website handling). It has been discussed > previously, but no solution has been proposed as of yet. I am quite sure we had this behaviour since forever. Did you verify that this is not old? regards, Lukas --- ## #26245 — Glenn Richmond — 2006-10-25T09:35:56+00:00 URL: https://externals.io/message/26242#26245 · In reply to: #26243 Done, I've just tested on PHP Version 5.2.0RC7-dev (today's snapshot) and get the same result. The code is: `session_start()`; $_SESSION['testing'] = $contents; `session_write_close()`; unset($_SESSION['testing']); `session_start()`; $newcontents = $_SESSION['testing']; `session_write_close()`; There is a php file prepended to all code executions that sets up the custom handler and it works for the first session_start, but on the second call to session_start, I get: *Fatal error*: `session_start()` [function.session-start]: Failed to initialize storage module: user (path: /tmp/php_sessions) i.e. It still knows that it's a custom defined handler, but the references to the functions are gone. If I make a call to session_set_save_handler after the first session_write_close, the problem goes away. Of course, this isn't practical for a large number of sites. Glenn. Antony Dovgal wrote: > > > […] > > I don't think it makes any sense to ask such questions on the list, > for the answer in most cases is "they are doing fine, thank you". > Just grab the latest snapshot, test it and report any issues you see. > > > […] --- ## #26246 — Lukas Kahwe Smith — 2006-10-25T09:54:37+00:00 URL: https://externals.io/message/26242#26246 · In reply to: #26245 Glenn Richmond wrote: > i.e. It still knows that it's a custom defined handler, but the > references to the functions are gone. If I make a call to > session_set_save_handler after the first session_write_close, the > problem goes away. Of course, this isn't practical for a large number of > sites. And this worked in previous releases? I am very certain that this was the bahviour on PHP4, not sure about 5.x though. regards, Lukas --- ## #26247 — Glenn Richmond — 2006-10-25T11:20:16+00:00 URL: https://externals.io/message/26242#26247 · In reply to: #26246 No, it's never worked as far as I know, but it doesn't mean it's not a bug. The session functions shouldn't have to be reset every time session_write_close or session_destroy is called. It's been logged previously at: http://bugs.php.net/bug.php?id=32330. It's also a consistent issue in varoius forums of major PHP projects. Any idea on a solution for this one? Regards, Glenn. Lukas Kahwe Smith wrote: > Glenn Richmond wrote: > > > […] > > And this worked in previous releases? > I am very certain that this was the bahviour on PHP4, not sure about > 5.x though. > > regards, > Lukas --- ## #26248 — Lukas Kahwe Smith — 2006-10-25T11:38:08+00:00 URL: https://externals.io/message/26242#26248 · In reply to: #26247 Glenn Richmond wrote: > No, it's never worked as far as I know, but it doesn't mean it's not a > bug. The session functions shouldn't have to be reset every time > session_write_close or session_destroy is called. It's been logged > previously at: http://bugs.php.net/bug.php?id=32330. It's also a > consistent issue in varoius forums of major PHP projects. Ah ok, it sounded to me like you were reporting a BC break. I agree this should be fixed if possible, but its not something that will get addressed in 5.2.0 simply due to release management concerns. Hopefully for 5.2.1 someone comes up with a fix. regards, Lukas --- ## #26249 — Glenn Richmond — 2006-10-25T11:57:44+00:00 URL: https://externals.io/message/26242#26249 · In reply to: #26248 ok, sure. I'm a little more familiar with the php code base now than when I first looked at the bug a few weeks ago, so I'll have another look and see if I can come up with a fix in the mean time. Should it be re-logged as a bug to get it some more attention in the future? Regards, Glenn. Lukas Kahwe Smith wrote: > Glenn Richmond wrote: > > […] > > Ah ok, it sounded to me like you were reporting a BC break. I agree > this should be fixed if possible, but its not something that will get > addressed in 5.2.0 simply due to release management concerns. > > Hopefully for 5.2.1 someone comes up with a fix. > > regards, > Lukas --- ## #26250 — Lukas Kahwe Smith — 2006-10-25T12:16:36+00:00 URL: https://externals.io/message/26242#26250 · In reply to: #26249 Glenn Richmond wrote: > look and see if I can come up with a fix in the mean time. Should it be > re-logged as a bug to get it some more attention in the future? more bug reports are not going to help .. just append the existing one with a patch. regards, Lukas