Newest at the top
2025-03-16 21:48:21 +0100 | robertm | (robertm@lattice.rojoma.com) robertm |
2025-03-16 21:48:05 +0100 | joeyadams | (~joeyadams@syn-162-154-010-038.res.spectrum.com) (Quit: Leaving) |
2025-03-16 21:47:16 +0100 | <merijn> | No, just A Doctor :p |
2025-03-16 21:47:01 +0100 | <EvanR> | are you The Doctor now |
2025-03-16 21:46:41 +0100 | robertm | (robertm@lattice.rojoma.com) (Quit: ...) |
2025-03-16 21:46:38 +0100 | <merijn> | I just realised some libraries I wrote are now a decade old :O |
2025-03-16 21:46:23 +0100 | <merijn> | EvanR: I already got that part down |
2025-03-16 21:46:16 +0100 | <EvanR> | not yak hair per second xD |
2025-03-16 21:45:49 +0100 | <EvanR> | optimize for long term maintainability |
2025-03-16 21:45:31 +0100 | <joeyadams> | I'm the devil on your left shoulder. |
2025-03-16 21:44:53 +0100 | <merijn> | joeyadams: Your doing a lousy job at convincing me not to yak shave :p |
2025-03-16 21:44:03 +0100 | <joeyadams> | If CApiFFI is better, why not contribute that to direct-sqlite, then write your interface on top of Database.SQLite3.Bindings or Database.SQLite3.Direct ? |
2025-03-16 21:43:17 +0100 | <merijn> | joeyadams: Sure, but I think iff I do significant work on it, I probably just want to also do something like base it on ResourceT |
2025-03-16 21:42:29 +0100 | <joeyadams> | merijn: What if you simply rewrite the direct-sqlite FFI bindings to use CApiFFI? I did a lot of work on this module (10 years ago) to separate those out. |
2025-03-16 21:42:22 +0100 | <[exa]> | merijn: re time tracking, didn't some of these pomodoro timer tools do that? I recall one that actually asked what you're doing, which is for me WAY more reliable than trying to remember to run that single terminal command |
2025-03-16 21:41:31 +0100 | <EvanR> | are all problems equally weighted (in lieu of suspecting some of them don't really exist) |
2025-03-16 21:41:28 +0100 | <merijn> | EvanR: That last one isn't a problem and more a "description" of why it's not being used :p |
2025-03-16 21:40:51 +0100 | <merijn> | But by now it's been in GHC for over a decade so I think it's safe to say "screw anyone who can't work with it" :p |
2025-03-16 21:40:51 +0100 | <EvanR> | you're reporting so many problems now I'm starting to get incredulous |
2025-03-16 21:40:07 +0100 | <merijn> | joeyadams: CApiFFI is great, the main "problem" it has is that it was "too new" when direct-sqlite and stuff was actively hacked on |
2025-03-16 21:40:00 +0100 | <joeyadams> | Had an interesting chat with my boss (who has never heard of Haskell) why this mysterious program named "cabal" was flagging the virus scanner on my work laptop. |
2025-03-16 21:39:28 +0100 | <joeyadams> | CApiFFI seems interesting. One fun thing I learned about hsc2hs is that virus scanners *really* don't like programs that use every win32 function, such as the win32 package. |
2025-03-16 21:39:05 +0100 | <EvanR> | (the more commits there are the more accurate it gets) |
2025-03-16 21:38:43 +0100 | <merijn> | etc. is a single terminal command and no mouse, and then enrich/attribute that time to projects *later*. |
2025-03-16 21:38:37 +0100 | <merijn> | At any rate, my main problem is that most time tracking tools seem to "frontload" the organising/attribution of time, so switching tasks becomes very burdensome. Even more so when they're in custom applications or browsers which mean I have to *gasp* use the mouse. So I just want to build something that lets me quickly type a command and some free text labels about what I'm doing so starting, switching, |
2025-03-16 21:38:34 +0100 | <EvanR> | (a statistical filter that converts git logs into timesheets for billing sounds... just stupid enough to work) |
2025-03-16 21:38:25 +0100 | gawen | (~gawen@user/gawen) gawen |
2025-03-16 21:36:47 +0100 | gawen_ | (~gawen@user/gawen) (Ping timeout: 268 seconds) |
2025-03-16 21:36:45 +0100 | weary-traveler | (~user@user/user363627) user363627 |
2025-03-16 21:36:11 +0100 | <EvanR> | basically |
2025-03-16 21:36:06 +0100 | <merijn> | EvanR: Marry SQLite and have it's baby? |
2025-03-16 21:36:03 +0100 | <EvanR> | use SQLite (in C) |
2025-03-16 21:35:54 +0100 | <EvanR> | then I have a better idea |
2025-03-16 21:35:52 +0100 | <[exa]> | EvanR: ...then pipe through GPT to make the 350h look like a LLM glitch, all done |
2025-03-16 21:35:42 +0100 | <merijn> | EvanR: Yes, at SQLite itself, which is glorious and without fault and I love it uncondtionally :p |
2025-03-16 21:35:20 +0100 | <EvanR> | does this chain of "The problem is" terminate? |
2025-03-16 21:34:49 +0100 | <EvanR> | and log 350 hours a day of work |
2025-03-16 21:34:47 +0100 | <merijn> | EvanR: The problem is that it uses hsc, which is annoying to work with, and it seems like you could outsource 95% of the work to CApiFFI and I'd rather maintain that :p |
2025-03-16 21:34:34 +0100 | <EvanR> | take the git commit timestamp and interpolate an exponential decaying tail of time leading up to it, and integrate |
2025-03-16 21:33:58 +0100 | <merijn> | [exa]: Well, it's more that I am terrible at tracking my time spent on various projects, meaning it's a nightmware to write hours on projects at the end of the day :p |
2025-03-16 21:33:55 +0100 | <EvanR> | make it keep up aggressively |
2025-03-16 21:33:23 +0100 | <[exa]> | merijn: I still read that it's technically FOR WORK |
2025-03-16 21:33:18 +0100 | <merijn> | EvanR: The problem is that it's not keeping up aggressively enough with SQLite :) |
2025-03-16 21:33:10 +0100 | <EvanR> | if the problem is the library functionality itself, maybe fix the library |
2025-03-16 21:32:57 +0100 | <EvanR> | if the problem is the API, maybe make a wrapper |
2025-03-16 21:32:52 +0100 | <merijn> | [exa]: So rebuilding the entire Haskell SQLite stack from first-principles, while great in principle, is not conducive to me getting a working solution to my actual problem :p |
2025-03-16 21:32:22 +0100 | <merijn> | [exa]: The problem is that I wanted to hack together something quick to solve a work problem, but since it's not a FOR WORK problem I can't do it on company time |
2025-03-16 21:32:09 +0100 | <geekosaur> | file a bug for that, both the absence of the option and that it happens in the first place (hoogledbs are supposed to be local, nothing related to them should ever be in an sdist) |
2025-03-16 21:32:02 +0100 | <[exa]> | last time I did a rdf triple store out of pure urge to shave that |
2025-03-16 21:31:35 +0100 | <TMA> | merijn: contrariwise! doing something you don't exactly immediately need is the way of making progress... into The Great Swamp of Despair |