Fixes due to some inconsistencies in HHVM vs PHP #66
Conversation
| @@ -62,7 +62,7 @@ Test#2 { "0" : null } | |||
| OK | |||
| Test#3 { "0" : 123 } | |||
| OK | |||
| Test#4 { "0" : 4.23 } | |||
| Test#4 { "0" : 4.23%d } | |||
There was a problem hiding this comment.
AFAIK, this shouldn't be necessary if you bump libbson. Older versions printed more precision than was necessary.
Also, %d doesn't have a "zero or more" variant (like %s and %S), so this could fail for newer libbson versions that only print 4.23, no?
There was a problem hiding this comment.
Oh, perhaps. Hadn't realize this was a libbson version. Will try that.
HHVM has the __pclass property last, and not first. That doesn't matter functionality wise, but it impacts the hexdump.
|
Two failures for PHP 5.5 and 5.6:
That should be resolved with #64. |
|
|
||
| ?> | ||
| ===DONE=== | ||
| <?php exit(0); ?> |
There was a problem hiding this comment.
IIRC, we discussed ===DONE=== previously and it should stick around.
|
This change seems to have broken travis: ========DIFF======== 001+ Fatal error: Cannot use object of type MyClass as array in /home/travis/build/10gen-labs/mongo-php-driver-prototype/tests/bson/bug0334.php on line 20 001- string(7) "MyClass" ========DONE======== FAIL PHPC-334: Injected __pclass should override a __pclass key in bsonSerialize() return value [tests/bson/bug0334.phpt] |
|
@bjori: Correct. I mentioned that in #66 (comment). |
No description provided.