| 1 |
On Wed, 2023-01-18 at 20:48 -0500, Joshua Kinard wrote: |
| 2 |
> |
| 3 |
> So is adding a default definition of TZ to our base system |
| 4 |
> /etc/profile something we want to look at? I |
| 5 |
> haven't tried any other methods of benchmarking to see if not making |
| 6 |
> those additional syscalls is just placebo |
| 7 |
> or if there are actual impacts. Given how long this oddity has been |
| 8 |
> around, I can't tell if it's a genuine |
| 9 |
> bug in glibc, an unoptimized corner case, or just a big |
| 10 |
> nothingburger. |
| 11 |
> |
| 12 |
|
| 13 |
I thought about doing this on my laptop, and talked myself out of it. |
| 14 |
The main counter-arguments are, |
| 15 |
|
| 16 |
1. ICU doesn't handle the :/etc/localtime format at the moment, |
| 17 |
|
| 18 |
* https://unicode-org.atlassian.net/browse/ICU-13694 |
| 19 |
* https://github.com/nodejs/node/issues/37271 |
| 20 |
|
| 21 |
You could readlink() it or whatever at boot, but that will cause |
| 22 |
changes to /etc/localtime to be mysteriously ignored. |
| 23 |
|
| 24 |
2. The stats are there for a "good" reason, namely to let glibc |
| 25 |
know if the timezone has changed on the fly. |
| 26 |
|
| 27 |
The first one is only a temporary deal-breaker, but the second is a |
| 28 |
tradeoff involving how often your timezone changes (user-dependent) and |
| 29 |
what the real performance impact is (probably not much). |
Archives