#872 Consider changing function naming recommendations in style guide a day ago

Comment by ~sircmpwn on ~sircmpwn/hare

An example of the subject/verb/object ordering issue is hare::module:: and the build driver using the style finish_* and free_*, whereas most other modules use *_finish and *_free.

FWIW the build driver is wrong here, it should be *_finish.

We should normalize (and document in the style guide) using "finish" for freeing resources in values, and "free" for freeing pointers and slices

https://harelang.org/documentation/usage/conventions.html#verbs-for-allocation-strategies

bufio:: has both init and newscanner, so even within the same module there's inconsistency in initialization functions. Also, as noted in the original ticket, there's encoding::utf8::decode vs base64::newdecoder.

This is not necessarily a problem per-se, if one feature of a module is its "flagship" or so then it's fine for it to use an abbreviated form while the ancillary features have longer names.

We copy the names of C functions in some places, but most of the time the function does something different than the C function with the same name, so copying the name is misleading:

Also FWIW copying C names in order to hint at what something does by leaning on the user's prior knowledge of C (if applicable) is a good idea imo, even if the functions differ slightly. See fnmatch, os::chmod, etc.

On the topic of strerror: its name doesn't match the current convention for other similar functions

It does: it strings an error. str is a verb.

#290 Empty emails are sent if the gpg agent fails to unlock the key 12 days ago

Ticket created by ~sircmpwn on ~rjarry/aerc

If the user mistypes their password or cancels the agent operation then aerc will send an empty email before bouncing back to the review screen for another attempt. It shouldn't do that.

#10 Free/libre assets 16 days ago

Ticket created by ~sircmpwn on ~sircmpwn/betamine

TrueCraft's assets are libre but I don't especially dig the vibe

#9 Improve chunk scheduler from O(n²) 16 days ago

Comment by ~sircmpwn on ~sircmpwn/betamine

Chunk scheduler was removed in favor of letting the server manage chunks

REPORTED RESOLVED WONT_FIX

#327 Debian repository signature has expired 22 days ago

Comment by ~sircmpwn on ~sircmpwn/sr.ht

REPORTED RESOLVED NOT_OUR_BUG

#329 unable to add gpg key 22 days ago

Comment by ~sircmpwn on ~sircmpwn/sr.ht

Send your question to the sr.ht-discuss mailing list. This bug tracker is not for end-user support.

REPORTED RESOLVED INVALID

#9 Improve chunk scheduler from O(n²) 29 days ago

Ticket created by ~sircmpwn on ~sircmpwn/betamine

It can probably just be O(n)

#8 Various optimization ideas a month ago

Ticket created by ~sircmpwn on ~sircmpwn/betamine

• ✓ Make chunk frustrum AABBs smaller by limiting them to the height of the chunk
• Split chunks into 16x16x16 blocks to (1) speed up tesselation and (2) improve frustrum culling
• Reduce size of vertex attributes and offload more logic to shaders
• Move tesselation off of the main thread

#7 Defer chunk unload by LRU a month ago

Comment by ~sircmpwn on ~sircmpwn/betamine

REPORTED RESOLVED CLOSED

#7 Defer chunk unload by LRU a month ago

Ticket created by ~sircmpwn on ~sircmpwn/betamine

To avoid thrashing the tesselator if the user happens to be wandering around a small area that overlaps chunk boundaries.