2d9885c introduced some regressions. This addresses those.
* Don't throw return type notice or session write failure when in an exception
* Fix tests to properly return true/false since null is no longer falsy/successy
* Rerecord a few tests to accomodate difference in raised warnings
According to the documentation, returning TRUE from
user based session handlers should indicate success,
while returning FALSE should indicate failure.
The existing logic relied on casting the return value
to an integer and returning that from the function.
However, the internal handlers use SUCCESS/FAILURE
where SUCCESS == 0, and FAILURE == -1, so the following
behavior map occurs:
return false; => return 0; => return SUCCESS
return true; => return 1; => return <undefined>
Since the session API checks against FAILURE,
both boolean responses wind up appearing like "not FAILURE".
This diff reasserts boolean responses to behave as
documented and introduces some special handling
for integer responses of 0 and -1 so that code can be
written for older and newer versions of PHP.
Allows user session handlers to create session IDs by adding an optional
7th argument to session_set_save_handler() and a create_sid() method
to SessionHandler.
A lot of code already existed to allow a custom create_sid handler, but
lacked a specific implementation.
Therefore I have added a 7th (optional) argument
session_set_save_handler, to allow a user function to be supplied for
session id generation.
If a create_sid function is not supplied, the default function is
called in its absence to preserve backwards compatibility.
Likewise create_sid only added to SessionHandler class, and not the
interface to maintain backwards compatibility. If the result is not
overridden, the default is called.
I don't have time right now to leave in the good ones and remove
only the bad ones.
There are some semantical changes which I reject, because
they aim at fixing a bug which is at a completely other location.
Then SID does not gefined anymore properly. (This broken patch
has not been sent to me at all.)
Also, there were *so* many whitespace changes which already
make these commits bogus.
- Move to the new ts_allocate_id() API
This patch is *bound* to break some files, as I must have had typos somewhere.
If you use any uncommon extension, please try to build it...
script writers to release the lock associated with the session lock
before the request finishes.
You can pass arrays now to session_set_save_handler(), so that the handlers
can be located in an object for better abstraction.
