| 20 Jan 2025 |
 rrix | oh heck it still fails | 04:20:13 |
| rrix | oh jeez my branch was out of date!!! | 04:21:39 |
| rrix | i didn't remove update_release | 04:21:48 |
 adisbladis | I just pushed that straight to master | 04:22:03 |
| rrix | cheers | 04:22:07 |
| rrix | haha | 04:22:08 |
|  cyphix 🇿 joined the room. | 04:45:28 |
|  @lambda-mike left the room. | 10:22:02 |
|  @liff:matrix.org left the room. | 11:37:17 |
 ge24z009 | It is very hard to compare exactly though because my Emacs runs pretty fast already and I don't have the many GC pauses normally.
The old GC could also be deferred by raising thresholds and that is not possible anymore, but my startup only changed from 0.25 to 0.35 so that seems good.
I tried to use this benchmark package but it makes use of manually collecting garbage before each timer start...
On some benchmarks, like running rec Fibonacci to 40, the igc branch was slightly faster than master. I don't think thats too meaningful as a benchmark either...
Vibes based, it feels good and I think I will keep using it!
Memory usage is higher for sure.
| 19:59:50 |
| ge24z009 | * It is very hard to compare exactly because my Emacs runs pretty fast already and I don't have the many GC pauses normally.
The old GC could also be deferred by raising thresholds and that is not possible anymore, but my (emacs-init-time) only changed from 0.25 to 0.35 so that seems good.
I tried to use this benchmark package but it makes use of manually collecting garbage before each timer start...
On some benchmarks, like running rec Fibonacci to 40, the igc branch was ever so slightly faster than master. I don't think thats too meaningful as a benchmark either because there are so many different GC scenarios.
Vibes based, it feels good and I think I will keep using it :)
Memory usage is higher for sure though.
| 20:01:38 |
 João Moreira | ge24z009 Thank you for your detailed explanation! So basically there's no gc-cons-threshold and gc-cons-percentage in igc, or no need for setting that up? I'll be trying it out specially if it gets into emacs-overlay. Also what are the specific things you do in order to get "(emacs-init-time) only changed from 0.25 to 0.35". Mine takes some seconds.. | 20:49:26 |
| João Moreira | * ge24z009 Thank you for your detailed explanation! So basically there's no gc-cons-threshold and gc-cons-percentage in igc, or no need for setting that up? I'll be trying it out specially if it gets into emacs-overlay. Also what are the specific things you do in order to get "(emacs-init-time) only changed from 0.25 to 0.35"? Mine takes some seconds... | 20:49:52 |
| ge24z009 | Yes, they have 0 effect when using the mps feature. I submitted a pull request for Emacs overlay yesterday too! My PR will probably need to be changed to use Savannah instead of Emacs mirror later though (when it is not broken anymore).
| 20:52:52 |
| ge24z009 | Basically alot of use-package with :hook after-init.
The stuff which changes the first frame is mostly in early-init + Some doom performance hacks | 20:55:00 |
| ge24z009 | But right now my Emacs doesn't really work bc I deleted many packages while configuring and now everything is down... | 20:55:42 |
| 21 Jan 2025 |
|  lazyschildpad joined the room. | 17:10:01 |
| rrix | zebrag: i wasn't expecting to get called "spammer" today in the room i admin ... 🤷 i think you should keep sharpening the skills and learning about emacs and nix | 18:14:54 |
 zebrag | rrix: It's very strange because it was a very productive session. Some people really have nothing to do but be bullies. And one of them is admin in the room. | 19:35:05 |
| rrix | i think i'll talk about it in the mod backchannel chat with folks later | 19:36:05 |
| rrix | but for now this room can handle nix and emacs just fine, we don't only complain when the overlay is broken ;) | 19:36:26 |
| zebrag | Yeah, I know, but the other room is fine too, at least in the description of the room. And as chinmay said, yeah you can scroll past the code you are not interested in. And no, it's not irc, yeah, we can paste code in it, there are all those pretty-print features to encourage us doing that. And because the code is visible directly in the thread of the conversation we can easily come back to it later. And the thing we pasted are of general interest for emacs user. | 19:42:45 |
| zebrag | So I really don't feel guilty, but I fell scared | 19:43:10 |
| zebrag | Maybe not scared, but it's very uncomfortable | 19:44:20 |
| zebrag | rrix: Have you seen my code? It was good! I'm quite proud of that solution :) | 19:45:26 |
| rrix | definitely understand and feel similar | 19:46:01 |
| rrix | why don't you paste your code in here ;) | 19:46:11 |
| 22 Jan 2025 |
| rrix | hm so there were reports in https://github.com/nix-community/emacs-overlay/issues/460 that the builds themselves of emacs-pgtk etc are failing because the fetcher tries to get the tarballs from cgit 🤦 | 06:41:59 |
| rrix | i'm kind of surprised there's no one in nixpkgs issue tracker noticing that fetchFromSavannah builds will be broken... | 06:45:16 |
| rrix | i guess that's me | 06:45:21 |
