AndreasPK's Avatar

AndreasPK

@andreaspk.bsky.social

Compiler enthusiast, GHC Maintainer, working with Well-Typed.

24 Followers  |  11 Following  |  16 Posts  |  Joined: 13.12.2024  |  1.7086

Latest posts by andreaspk.bsky.social on Bluesky

> 5% is a little excessive, IMHO. I can't imagine the Canadian prairies being 5% wildflowers instead of farmland.

What makes it seem excessive to you? Imo in the end pollinators need enough space to maintain a stable population. If that takes 5% then it takes 5%.

08.08.2025 08:57 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Post image

The debugger taunting me when I'm hungry.

28.07.2025 12:42 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

Finally I learned what upgrading python projects is like.

03.06.2025 12:31 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Programmers are usually fed a steady diet of features and bug fixes. But occasionally they get to work on performance problems. This development methodology is known as intermittent fasting.

21.05.2025 16:19 โ€” ๐Ÿ‘ 86    ๐Ÿ” 9    ๐Ÿ’ฌ 4    ๐Ÿ“Œ 0

Today I came across one of my comments on someones blog from 14 years ago. It was peak denglisch.

07.05.2025 23:25 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

And if it goes wrong you can still pivot to bread theory.

25.04.2025 00:42 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

And a few hours later ...

16.04.2025 18:46 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image 16.04.2025 12:09 โ€” ๐Ÿ‘ 3    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Andreas Klebingers Blog - Unknown functions and their arity.

Some musings about unknown function calls and arity tracking in GHC:
andreaspk.github.io/posts/2025-0...

Also: Five years between blog posts! Time flies when Covid happens.

12.04.2025 21:50 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

Who doesn't love consistency?

11.04.2025 16:51 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image Post image

Benchmarked haskell a package I maintain across some GHC versions.

Surprised by the degree of improvement.

10.04.2025 13:45 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

If you haven't thought about unification before probably yes.

07.04.2025 19:01 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Lua. Because everyone should experience 1-indexed arrays at least once.

07.04.2025 18:59 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
stageNumber stage = stage+1

stageNumber stage = stage+1

Things that make me hate my job sometimes.

07.04.2025 18:54 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Hackage Search by Serokell

It's not perfect but you could try using the serokell hackage code search to search for uses of those: hackage-search.serokell.io

20.12.2024 18:46 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

As far as I know if one needs ($label | 3) or similar often you can still avoid computing this at runtime by using linker scripts which will compute the relevant address at runtime and expose it as a symbol. But I haven't had to use that myself yet.

13.12.2024 10:51 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

It comes down to what the linker relocations support and that is in the end "arbitrary".

Offsetting an address is a common operation (read a field in a struct), and so is multiplication (indexing into an array). So they have good support. By comparison ($label | 3) is pretty exotic.

13.12.2024 10:47 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

@andreaspk is following 11 prominent accounts