|
| From: | Paul Eggert |
| Subject: | bug#49340: small sort takes hours for UTF-8 locale |
| Date: | Fri, 2 Jul 2021 17:25:51 -0700 |
| User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.11.0 |
On 7/2/21 4:19 PM, Pádraig Brady wrote:
we might be able to improve things. For example, using strxfrm() + strcmp() to minimize processing.
I tried that long ago, and it was waaayyy slower than strcoll in the typical case. glibc strxfrm is not at all optimized.
Which is fine, since strxfrm is a dumb API: its only point is performance but its portable API is inherently low-performance for typical uses. I've never seen it useful.
In short, this is a glibc strcoll bug and should be fixed there.
| [Prev in Thread] | Current Thread | [Next in Thread] |