18 Mar 2024 |
notafile | yeah, sounds a bit annoying to fix, maybe I'll have a look at it again | 13:03:37 |
utf-4096 | i noticed issue \#498 is still open, i think it might be better to instead use a proper VXL handling library such as libvxl (by bytebit) or libmapvxl (by darkneutrino, used by spadesx) since the current vxl_c code uses global variables which aren't great either | 13:09:06 |
notafile | yeah that would be very much appreciated | 13:10:10 |
notafile | either one works for me | 13:10:28 |
utf-4096 | i'll probably try to implement both to figure out which one's the fastest | 13:11:34 |
notafile | I don't think it hugely matters, with how bad the map data structure is | 13:13:19 |
notafile | unless you want to use it for the in-memory representation too | 13:14:04 |
notafile | in which case I feel like either is probably gonna be faster than what we have right now? | 13:14:54 |
notafile | the thing I didn't like about libvxl last time I looked at it is that it had zero bounds checking and any corruption in the file was just completely undebuggable | 13:16:21 |
utf-4096 | both of them have practically the same api, it wouldn't be hard to make bindings for both of them | 13:16:30 |
utf-4096 | * both of them have practically the same api, it wouldn't be hard to make bindings for both | 13:16:37 |
utf-4096 | In reply to@notafile:matrix.org the thing I didn't like about libvxl last time I looked at it is that it had zero bounds checking and any corruption in the file was just completely undebuggable libmapvxl suffers from the same thing | 13:16:44 |
utf-4096 | well kind of | 13:16:54 |
utf-4096 | it gives you a warning but that's it | 13:17:10 |
utf-4096 | though if the map is corrupted, when it tries to load it it messes up and eventually segfaults | 13:17:26 |
utf-4096 | in my libmapvxl wrapper i did bound checks myself, on the cython side https://github.com/utf-4096/pymapvxl/blob/main/bindings/libmapvxl.pyx | 13:20:23 |
19 Mar 2024 |
utf-4096 | master's down (again!), which is a great reminder to dust off the multiple masters pr | 20:21:01 |
utf-4096 | (to myself) | 20:21:15 |
22 Mar 2024 |
utf-4096 | In reply to@utf-4096:matrix.org but i want to know if thats the proper solution or if this is an entire math rabbit hole @notafile ive put this change in the aloha test servers and it doesn't seem to cause any issues, do you know more about this? if there's no issue reverting the change am i ok to make a pr for regular pique? | 23:13:24 |
23 Mar 2024 |
notafile | sounds good | 08:23:22 |
notafile | I remember there was some other issue where just a straight cast was wrong | 08:24:01 |
utf-4096 | the ftol change was actually made by mat^2 so yeah i expected it to be accurate but it seems to break more things than it solves | 09:38:30 |
utf-4096 | we only recently switched to pique for aloha (been on pysnip for ages) and pysnip didnt have that change, and i dont think we've seen anyone complain about inaccuracies so 🤷 | 09:39:31 |
28 Mar 2024 |
utf-4096 | master is down... again | 17:19:09 |
utf-4096 | my shitty bash client can still do hellolan though :^) | 17:23:44 |
utf-4096 | Download image.png | 17:23:44 |
29 Mar 2024 |
| popey changed their profile picture. | 13:16:10 |
31 Mar 2024 |
sByte | In reply to @utf-4096:matrix.org sent an image. How much time that takes to render? | 15:48:07 |
utf-4096 | around 15s | 16:24:25 |
utf-4096 | would be much faster if i made it multithreaded but cba | 16:24:52 |