#dywypi

[eeveedev]

ok i totally forgot to do this haha

i haven't been doing nearly as much as i'd like honestly; all the things i want to do are big and huge and complicated, and when that happens i get intimidated by possibility and sort of freeze up

BUT

i pushed some dywypi work and have been gradually wrapping my head around where i want it to go again

i got flax (my roguelike, take four) sorta working how i want, though there are no other actors yet

i'm working on floraverse (spline), which is the most daunting of all, but i have the wiki basically sort of functional which is really nice to see.

i did a little boring administrative work with sanpera, trying to figure out wtf is up with a problem weasyl is having

i have a zillion things i want to do with spline, including come up with a way to make deploying and upgrading it turnkey, but these are kind of hard problems and the whole point of spline is that i'm really fucking picky about the architecture. which sucks, because of course, the architecture is crap until i figure out how a feature is going to work. it's all very paralyzing. which is annoying. and defeating. ughh.

anyway. yeah. workin on it.

[eeveedev]

uhh so i spent the last few hours finishing up this: https://github.com/wardi/urwid/pull/67

which is an attempt at making it easier to plug this python console UI library into async event loops

which is a whole lotta yak-shaving originally intended for my irc bot, dywypi

which i will actually work on directly again, someday.

[eeveedev]

uh eading uhhh

i got persona login half-working for floraverse, which is at the bottom of some yak-shaved pile of dependencies for making the wiki usable

i have something of an interest in openid connect as no-js fallback (even if persona does end up with a declarative flow, there's still no easy way to use it programmatically, and there's no native browser support anywhere) but openid connect appears to be some colossal enterprisey clusterfuck that things a stack of eight legalese spec documents is an acceptable way to explain anything

in fact reading about openid connect might be the reason i still feel so drained now, three days later, because i've done basically fuck-all all week

i keep tinkering with dywypi as well but it's gotten to the point where the next improvements to the core aren't really obvious without writing an interesting plugin, and writing an interesting plugin is a giant pain in the ass until i do a whole bunch of intricate things to the core, and that totally sucks. i wanted to make uno work for mel but i look at it and just make this gutteral gurgling noise. likewise i want to replace the terrible node.js thing we have at work that goes down all the time, but it's missing a bunch of features that are just fucking hard

and i haven't been writing or drawing or japaneseing either. so, not a successful week thusfar.

also i leave for pycon like a week for today which means an international flight over the weekend and etc so oh boy. it'll be fun but the dread is starting to set in. here's hoping i do anything useful this weekend

[eeveedev]

Do you have setup docs for dywypi anywhere? The standard filesharing bots are mostly terrible and I want to fix this. (Is there, BTW, a reason it doesn't do DCC beyond 'haven't gotten around to it yet'?)

there's a README; basially run it with python -m and away you go

it's not exactly super configurable yet  :)

also yes that's pretty much it.  IRC itself is hard enough and DCC is an entire separate protocol

[eeveedev]

i've been working on dywypi mostly; i would love for it to become a real useful thing but it needs a lot of good plumbing

i've been toying with the urwid integration too because more than anything i would looove to have a neat little local shell. i actually got this working so well that i could serve multiple terminal apps from the same asyncio process to multiple clients. kind of amazing.

the unfortunate bit is how much of urwid i had to hack around to make this work; i grabbed a copy of it and have been hacking at it most of tonight trying to make it work better, because i'm an idiot and want to shave every yak

i really need to work on spline

and do a zillion other things

[eeveedev]

i was really worn out all day today

fucked around with pelican a bit, looking at porting my blog to it; seems cool, just need to fix some octopress-specific features and actually write a theme ho ho...

also kicked ledger a bit without much success. accounting is hard. i don't know if i should even bother with this tbh since vanguard works well enough for a pretty graph and there's a lot of tedious keeping-up-to-date involved

uh

i went through workflowy and turned some bullet points into actual tickets on a few projects. which is nice. i always have a lot of stuff in the air though and i would reeeally like to get something more closer to released and out the door. it often gets to me that i do a lot of work and then sit on it and no one really benefits or even sees it.

so i'm going to try to make that a bit higher of a priority in the near future. dywypi and sanpera are closest to something resembling a release, i think; they're still far from being fleshed-out stable products, but they are at least things other people could make use of in the near future, maybe.

i also want to fucking blog something every time i make a release, because surely there will be something interesting to talk about in it.

i feel like i have a bit of a problem with knowing what to ever work on, too; i see pyscss now has its fair share of bugs and could use some more love, but that's yet another thing on my plate. how do i balance the time? or do i? does it actually matter as long as i keep moving forwards on things that interest me? maybe not

vyvanse sure fucking helps a lot though! even my goofing off is productive. and now i'm gonna lie in bed and hack dywypi.

[eeveedev]

heyoo

i spent the last week in SF for a yelp hackathon, though that only spanned the last two days

other people wanted to write some mafia-style irc games, and they wanted to use my bot as a base for it, so i joined their team and just hacked on the infrastructure

it came out as a great success i think

the bot base is dywypi and it is pretty cool; entirely asynchronous python 3. out of the box it doesn't do a whole lot yet since i was trying to clean up the plumbing but apparently it's easy enough to add stuff to

one of the neatest things imo is that it has a terminal interface, so you can just run it locally without actually connecting to irc at all. it's, uh, kinda super fucking glitchy atm, and is a port of crazy twisted code that was itself a combination of other people's crazy twisted code... but it's a good start. and it avoids the common pitfall of trying to make an auto-reloading bot, which never really works well.

it needs a lot of work alas but i'm hopin to spruce it up a bit soon and have it maybe replace the existing dywpi