#17129 closed enhancement (fixed)
Speed up is_serialized_string()
| Reported by: |
|
Owned by: |
|
|---|---|---|---|
| Milestone: | 3.2 | Priority: | normal |
| Severity: | minor | Version: | |
| Component: | Performance | Keywords: | reporter-feedback |
| Focuses: | Cc: |
Description
is_serialized_string() uses an expensive regular expression. It should use substring peeking like is_serialized() does. In my tests, this offers a time reduction of about 26%.
Attachments (2)
Change History (9)
#2
@
14 years ago
- Owner set to markjaquith
- Resolution set to fixed
- Status changed from new to closed
In [17779]:
#4
@
14 years ago
- Keywords reporter-feedback added
When I reviewed the changeset this is what came into my mind. I have not benchmarked so far if it's actually faster than the if / else approach.
#6
@
14 years ago
A note of caution: is_serialized_string() will return true for cases is_serialized() will return false. So both functions are not compatible with each other.
Related: #17375
As is_serialized_string() is used in core only after if(is_serialized()) has been true, and it's not same stage compatible, it could be even written much smaller by just checking for s/S as the first letter.
Note: See
TracTickets for help on using
tickets.
Edit: Oops, Ignore What I've said...
the last elseif can be changed to
elseif ( '"' !== $data[4] || '"' !== $data[$length-2] ) return false;