chrsig

the narrow window view is appreciated given the increased indent level of your comment

> You can see in chats 11 to 14 that we're entering the slop zone. The code the agent created has a critical bug, and it's absolutely failing to fix it. And I have no idea how to fix it, either.

This definitely relaxes my ai-hype anxiety

to be useful to whom, exactly?

w.r.t. your edit: Is there anything the community at large can do to aid your efforts?

You're sort of answering your own question. It was a matter of proximity. The thousand others were a greater distance in the initial conditions.

valgrind finds sooooo many more problems than just memory leaks

uninitialized memory, illegal writes, etc... There's a lot of good stuff that could be discovered.

I'm glad to see rsc still actively involved. And commenting on commit messages.

The older I get the more I value commit messages. It's too easy to just leave a message like "adding valgrind support", which isn't very useful to future readers doing archaeology.

Useful for people -- believe it or not not all of us internalize the language we use as part of our identity (and species).

From someone that grew up in New England, but experienced a 4.x in california: I assure you it does not feel small.

The revelation that the ground does not stay where I left it was quite disturbing.

I'm coming to expect that most things you see on the internet are written by an LLM. I can't wait for all the de-LLM denoiser products that're sure to come out combating llm spam.

Thanks for bringing up therapy. I've been in therapy for some time, which lead to leaving my former position. I agree with you that my brain is certainly more buggy than my side projects :)

Think of it like emacs. Browsers are perfectly good operating systems just needing a better way to view the web.

I've definitely done that as well. I generally don't have a problem with it, but it makes for an extra conversation the first time someone sees the pattern.

Given how some of the other ergonomic changes in go have gone (closures capturing loop variables, for instance), I'd support a change to lexical scoped defers if it were on the table.

yeah, I've always just extracted the loop body into a new function as a result

>Those who listened correctly predicted a whole bunch of things and were called paranoid by enablers.

was being called paranoid not also predictable?

i don't like unexpected things happening. they cause me to panic. i can expect errors.

maybe spend the time talking to his voters?

This is a great perspective on pedantry

how nested is the data in the parquet files?

i recall some of the same speculation. in my case we wound up getting a satellite connection, with a wonderful 600ms round trip time.

ah, good ol' capping out at 26.4kb/s...until 2005

i don't think they have been for a while.

you're not wrong, but i don't think that you're presenting the argument in a way that is going to be well received.

hopefully both gccgo and any llvm backed implementations eventually mature to production grade. I think the thing that'll hold them back is that the toolchain is (by definition) completely different. 'go build .' is pretty nice.

the biggest value they bring is that unclear parts of the specification=be brought to light and clarified.

this was actually one of the things that really turned things around for me actually. it took probably 25 years in between.

I a missing component is a plan for recourse if the student doesn't take to the new pace/material.

That was my biggest problem, that and I wasn't actually on-board for the skip. Educators need to learn how to admit when they fucked up and learn how to improvise a new strategy.

I'll call out 3b1b and khan academy for me. Especially over covid. Made math fun again.

My middleschool principal thought it'd be a good idea to skip me over pre-algebra into alg 1.

Turns out that doesn't work great, and I still have confidence issues because I have a hard time remembering the properties of addition & multiplication by name. I know the rules.

no...they'll just let the seniors die.

> he can’t cut either one without Congress. What is he going to do? Not send checks to senior citizens?

cut it anyway then deal with the 'consequences' in court. obviously.

someone doesn't remember y2k.

> and autovectorization is brittle, writing inference or training in a way that relies on it is the last thing you want)

I'm curious if you could speak more to this? Is the concern that operations may get reordered?

> To put it in perspective, until 2021 Go was always passing the data on the stack on function calls. It has improved since then but the overall design aims to ensure common scenarios are fast (e.g. comparisons against string literals are unrolled) but once you venture outside that or if it's an optimization that requires more compiler complexity - Go is far less likely to employ it.

I agree with this assesment.

The individual operations in the repository (e.g., dot product) look like they could be autovectorized. I'm assuming they aren't because of the use of a slice. I'm mildly curious if it could be massaged into something autovectorized.

Most of my observations re: autovectorization in go have been on fixed sized vectors and matrices where SSE2 instructions are pretty readily available and loop unrolling is pretty simple.

I'm curious what it would produce with the matrix in a single slice rather than independent allocations. Not curious enough to start poking at it, just curious enough to ramble about it conversationally.