| From: | Heikki Linnakangas <hlinnaka(at)iki(dot)fi> |
|---|---|
| To: | Chao Li <li(dot)evan(dot)chao(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: Avoid rescanning prefixes in appendStringLiteralDQ() |
| Date: | 2026-11-19 11:43:41 |
| Message-ID: | 2584e9d9-5e5b-4ff7-a773-1f38037b69af@iki.fi |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On 19/11/2026 08:32, Chao Li wrote:
> While reviewing another patch, I spotted a trivial optimization in
> appendStringLiteralDQ().
>
> Looking at the code:
> ```
> /*
> * Make sure we choose a delimiter which (without the trailing $) is not
> * present in the string being quoted. We don't check with the trailing $
> * because a string ending in $foo must not be quoted with $foo$.
> */
> while (strstr(str, delimBuf->data) != NULL)
> {
> appendPQExpBufferChar(delimBuf, suffixes[nextchar++]);
> nextchar %= sizeof(suffixes) - 1;
> }
> ```
>
> The while loop always searches from the beginning of str. But
> actually, a later search can just start from the previously returned
> pointer. This function is currently only used by pg_dump. I don’t
> expect a noticeable performance improvement in typical use, but it
> seems to me to be a tiny but harmless optimization.
Makes sense. There's a similar loop in pg_get_functiondef(), too. I
applied the same optimization there, and pushed.
- Heikki
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Bertrand Drouvot | 2026-11-19 11:59:59 | Re: Report buffer access strategies in EXPLAIN |
| Previous Message | Nazir Bilal Yavuz | 2026-11-19 10:59:22 | Re: Switching XLog source from archive to streaming when primary available |