The initially observed issue was failing tests in our test suite. As I traced that through I eventually landed on a Rust function that returned a bool to Mercury code. Logging the value on the Rust side was fine, but then attempting to log the same value on the Mercury side was crashing the Mercury runtime. Eventually I logged the value from a C foreign_proc in Mercury and was seeing it was receiving the value 16 instead of 0 or 1. It was at this point that I suspected a compiler bug. Comparing the assembly of the function was also suspect:

1.99:

Disassembly of section .text.fontcode_rust_font_has_glyf:

0000000000000000 <fontcode_rust_font_has_glyf>:
       0: 0f b6 87 9f 01 00 00          movzbl  0x19f(%rdi), %eax
       7: c3                            retq


1.83:

Disassembly of section .text.fontcode_rust_font_has_glyf:

0000000000000000 <fontcode_rust_font_has_glyf>:
       0: 0f b6 87 9f 01 00 00          movzbl  0x19f(%rdi), %eax
       7: 24 01                         andb    $0x1, %al
       9: c3                            retq

From there I was able to build a small reproducer and open an issue on the rust repo.

score 1 · canonical

I found and reported the bool FFI miscompilation. The behaviour I was observing was very confusing, and I’m generally extremely hesitant to blame the compiler. Eventually I brought the problem to the rest of the team and started with “I’m running into some strange behaviour, which is smelling like a compiler bug.” somewhat reassuring that my suspicions did turn out to be correct in this case.

score 3 · canonical

indeed I would like that too, there usual way of submitting reports is through email on their support page.

Yes I sent them an email with a few queries.

score 1 · canonical

I too am a paying ente customer (200Gb plan) and am not happy about the ramping up of the inclusion of LLM generated code. They also seem to be spinning up new products every other month when the reason I chose them was solely photos. The experience of which still has plenty of room for improvement before diluting their focus with additional products. The most recent of which I’m dismayed to discover is “Ensu, our first step toward a private, personal LLM app that runs on your device and grows with you over time.”

They seem way too LLM brained for me. Zeitkapsl looks interesting, I’ll need to evaluate it further. One positive sign is they host their code on Codeberg, which doesn’t permit largely LLM generated projects. How has your experience been with it?

Edit: Not off to a great start. I tried building the desktop app as I’m on an uncommon distro and the Makefile references a target that seems to have been deleted 25 Sep: Makefile tries to run copy-models in ml but that Makefile was deleted in a commit from 25 Sep. I went to open an issue about it, but issues are not enabled on zeitkapsl/zeitkapsl…

score 1 · canonical

Thanks, I appreciate hearing that. I had fun building the site and seeing how far I could get with default styles. I also tested in Dillo and links as I figured if it worked there it would work pretty much anywhere.

score 1 · canonical

I run a Chimera mirror at https://au.mirror.7bit.org/. To ensure atomic updates I use the atomic-rsync script, which is included in Rsync tarballs, but is not currently part of the Chimera rsync package (the mirror server runs Chimera).

atmoic-rsync manages a pair of directories, syncing into the inactive one, then swapping a symlink when the sync is done so what’s returned by the web server switches instantly. atomic-rsync uses the --link-dest option to populate the tree being synced from the currently active one to avoid storing duplicates of unchanged data.

This has worked well, although occasionally rsync would get stuck, or take a long time and the syncs would start overlapping. That was solved by invoking it with flock. My full cron line is:

30 *  *   *   *   chimera flock -w 59 /var/run/lock/chimera-rsync.lock /usr/local/bin/atomic-rsync -ah rsync://repo.chimera-linux.org/chimera/ /home/chimera/www/chimera/ && date -Iseconds > /var/www/chimera-sync.txt

When I set this up I wasn’t aware of the mechanism q66 used on the Buildbot workers, if I’d known that then I could have done the same.

score 3 · canonical

Thanks. I did expect it to last longer than this when I publicly announced the project, but it was still a good learning exercise and that’s a good outcome too.

score 2 · canonical