| From: | Heikki Linnakangas <hlinnaka(at)iki(dot)fi> |
|---|---|
| To: | John Naylor <jcnaylor(at)gmail(dot)com> |
| Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Speeding up text_position_next with multibyte encodings |
| Date: | 2018-12-23 00:45:05 |
| Message-ID: | 505c447a-5e2a-2c0d-becc-c9bd2a7ebc2f@iki.fi |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On 23/12/2018 02:32, Heikki Linnakangas wrote:
> On 23/12/2018 02:28, Heikki Linnakangas wrote:
>> On 14/12/2018 23:40, John Naylor wrote:
>>> I just noticed that the contrib/citext test fails. I've set the status
>>> to waiting on author.
>>
>> Hmm, it works for me. What failure did you see?
>
> Never mind, I'm seeing it now, with assertions enabled. Thanks, I'll
> investigate!
The bug was in handling empty inputs. text_position_setup assumed and
asserted that neither the needle nor haystack are empty, expecting the
callers to have handled those special cases already, but not all callers
did. Here is a fixed version.
- Heikki
| Attachment | Content-Type | Size |
|---|---|---|
| 0001-Use-single-byte-Boyer-Moore-Horspool-search-even-wit-2.patch | text/x-patch | 24.6 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Jeff Janes | 2018-12-23 01:49:58 | Make relcache init write errors not be fatal |
| Previous Message | Michael Paquier | 2018-12-23 00:41:03 | Re: Offline enabling/disabling of data checksums |