[ a loud jason spaceman song either about falling in love or struggling with heroin grows in volume to a crescendo ] yeah i'm good

[ philip glass's koyannisqatsi blasting through the door ] yeah i'm doing good thanks for checking in

this is probably my only contribution to computer science

i wonder if the editor of the posix standard is "ha. shows you. never had to implement leap seconds"

remembering the old trick of finding a loop in a linked list: free every item in turn and if there's a segfault, there's a loop

my record for "number of plan9 users i've been in the same conversation with" is still three, and so far, unbeaten

good morning

thinking about "an oral history of bank python" again

https://calpaterson.com/bank-python.html

The software engineer pipeline isn't a pipeline but rather a pulsejet. We are in the exhaust phase.

coming to the conclusion that i need more enrichment in my enclosure

@tay

the future of programming is a gigantic fruit machine pipeline and running down the isle pulling each lever in turn until you win, and people endlessly sharing strategies

so we can put a skin on a database, collect metrics, and show targeted gambling adverts. can't wait

sorry i read a very bad blog post about our new agentic future and the need for better sandboxing because the guy couldn't get the computer to stop deleting the code over and over and yeah i simply haven't been kicked in the head enough for this future

call me old fashioned but i don't really think the future of programming is per-request billing to defence contractors set on destroying the internet

or at least, a future we should aspire to

the thing about saying "one day they'll learn" is that the reason we have stories like "the hare and the tortoise" is that people don't

no-one loses on a scratch card once and thinks "i should stop here"

we have a whole "you lose 100% of the scratchcards you don't buy" mentality here

if people learned from consequence the world would be a much nicer place

like i have seen two blog posts about harnesses and at best it seems like people trying to paper over the cracks, a tacit admission that the tools cannot work without strict guidance, but the unkind reading is a gambler battling probability

the problem is, being heavily stochastic, it's really easy to trick yourself into believing that your nudges will work every time, or they have impact, or combine independently, and build superstitions

the code doesn't have to work every time, so

it's funny how i got told "you will be left behind" and as it turns out the skill you need to develop is simply spending more tokens when any problems occur

code not good? add more agents. spend more tokens

code still not good? stuff things in context window. call it a skill. spend more tokens

code still bad? break down the problem, spoonfeed computer line by line until code good, pretending you haven't been tricked tom sawyer style into doing the work and spending more tokens as a result

can't believe they've enshittified programming, thought we were a bit further down the list given the progress on other fronts

"why did you get into computers, tef?"

mostly by accident. i was more interested in mathematics as a teen, and ended up doing computer stuff at school as it was easy and let me avoid writing prose

"how did that work out, tef?"

well, my code never went anywhere, but my prose about code went viral a few times, and now my whole career is turning into writing prose for no-one to read, to build code no-one understands, so it's going well thanks for asking