Community
Chat Logs
Wednesday, July 8, 2026
- kcomhnall"call a method with all possible combinations" - reminds me of COBOL
- dreamrealthe problem with "call with all permutations" for me is that it relies on ... what for validation?
- dreamrealif the validation is consistent, that's good, I guess, but I usually want tests to test for failure as well
- ParaThe next inevitable steps will be generative testing and then fuzzing :)
- ParaI did get stuck pondering what would be the best way to do n-way cartesian product as a stream, though. I mean, if I did that, I'd want it to also try to be at least somewhat memory efficient as well.
- dreamrealgenerators a la python
- jreicherPara: You mean output as a stream, or formulate a notion of cartesian product for two streams as the "operands"?
- jreicherBecause I don't think all of the above can be streams.
- Parajreicher: n-ary Cartesian product as linear stream
- ParaWith executive decision on settling on Some™ strategy to choose how it is linearized.
- ParaSo not just two but, dunno, 17. And thus the need for memory efficiency.
- ParaThis is purely me entertaining myself, I have a few days left till vacation but all our supervisors and stakeholders are already on vacation and there's nothing left open in the backlog so have to fill the time somehow :D
- jreicherPara: so you'd be OK with the result being a stream but one of the "operands" being, say, an array?
- ParaI'm not sure what you mean.
- Para`Stream<List<T>> products = cartesianProducts(a1, a2, a,3 ...)` would be what I want, where T is probably nothing special, common type for the vararg arities provided or whatever.
- jreicherI wasn't sure if the a1, a2, a3, etc. were streams also
- ParaI don't think it's possible to produce a meaningful Cartesian product without knowing the bounds of the arities.
- dreamrealjust let the AIs guess, it'll be fine
- dreamrealhttps://www.reddit.com/r/java/comments/1uqiej1/replace_cc_with_java/
- cheeserOP sounds like an LLM
- ParaFriend of mine has found a new hobby from looking at Claude Code's open issues at GitHub.
- ParaIt's a glorious garbage fire.
- Shellit's really weird the last couple of years - you don't have any idea any more if any software is capable of doing what it says on the label for longer than a quick demo. claude code is just a leader there.
- dreamrealI was impressed by the string of non sequiturs!
- dmlloydI made another thing: https://github.com/smallrye/smallrye-serial
- nevetGitHub - smallrye/smallrye-serial: A safe and efficient Java serialization library
- javabotdmlloyd's title: "GitHub - smallrye/smallrye-serial: A safe and efficient Java serialization library · GitHub"
- dmlloydthe name is not exactly descriptive: it is an intermediate representation for serialized objects
- dmlloydso you can examine and manipulate the serialized form of objects
- dreamrealer... as usual your docs suck. Why, what are the default providers, what would the serialized form look like or what MIGHT it look like, how would I use this, how would I know if I needed or wanted this?
- ParaI should publish a microlib which has exactly one interface, AI. No content. That way you could just add "implements AI" everywhere.
- dreamrealdmlloyd: and what makes it efficient, dang it respond to me
- dmlloydwell, it's not inefficient
- dreamrealgrrrr
- dmlloydit just does the exact thing it says on the tin, that's all
- dreamrealit doesn't say WHAT IT DOES in a way that's uncompressed
- dreamrealI think YOU understand what it does well but to me you have not expressed it well in a way where I'm not having to spend a lot of cognition to understand it
- dmlloydit turns serialized Objects into Serialized objects
- dmlloydoo, I should make that be the description
- dreamrealyeah, that'll fix SO MUCH.
- dmlloydmy favorite feature though is if you feed some object into a Serializer, you can examine it using the Printer to see how it's put together
- dreamrealWell, you need to DESCRIBE that way more gooder. Maybe even with those weird things called "examples."
- dreamrealand MAYBE even "what can you learn from this, Mr. ordinary person? Why do you care? Here's why."
- dmlloydI'll pass your critique on to my associates, Claude and Bob
- dmlloydanyway it's not for normal plebes, it's for elite coderz only
- dreamrealI get that, but for the love of god... I mean, I want to write this up for BCN. Do you REALLY want me to answer all the questions I have for myself?
- dmlloydwhat do you think I pay you for?
- * dreamreal checks his bank account
- dreamreal... about what you get, apparently :D
- dmlloydheh, I almost missed this nugget: https://github.com/openjdk/jdk/pull/31441
- nevet8286300: Port JEP 425 to S390X by offamitkumar · Pull Request #31441 · openjdk/jdk
- javabotdmlloyd's title: "8286300: Port JEP 425 to S390X by offamitkumar · Pull Request #31441 · openjdk/jdk · GitHub"
- dmlloydvirtual threads on the mainframe
- Paramakes me think of this RISC-V GPU https://www.youtube.com/watch?v=qMR3IXF2sWw
- nevetYouTube: Home made GPU escalated quickly | bitluni
- dmlloydyeah I saw a reference to that in passing on Reddit recently, interesting idea
- cheeserhuh. TIL Aleph One, of "Smashing the Stack for Fun and Profit" fame, is now at Google.
- dreamrealI wonder if anyone's ever put a JVM on the E-series from unisys... 9-bit processors, y'all
- dreamrealthree octal sets make up a byte
- ParaWhat's the extra bit for? Or reasoning for it, rather?
- dreamrealthree octals
- ParaWith 3 bits spare? :P
- cheeserwhut?
- dreamrealerr... nine bit bytes = three octal sets.
- dreamrealno spare ANYTHING.
- Paraah right...but still why
- dreamrealBurroughs?
- dreamrealI mean, that's... really the answer
- dreamrealActually, the E-series was sperry, I guess.
- ParaMaybe it was a dare.
- dreamrealwith unisys you could never tell. Really decent hardware for what it was: a system 36 clone.
- dreamrealI did the e-series first, and went to night school on a system 36 - was very weird, because i was used to ECL and IPF and had to deal with "JCL" and "ISPF" instead at night. (I didn't finish the cert: they kept assigning things I knew how to do from work, so I did them like real programmers would, and they'd ding me for doing stuff they didn't teach me.)
- dreamreal"You can't use a quicksort! We haven't taught that! Bubble sort or bust!"
- dreamreal(not the actual problem but ... imagine that sort of thing. Over and over again.)
- dreamrealPara: found it. Army Signal Corps. They had six bits of signal as a requirement, and thus octal fit the data well; this lent itself to 36-bit words. The nine-bit byte came about from ascii support, as a natural widening, and this meant a word itself also did pretty well with number representation. This was all done before the standardization around 8-bit bytes.
- sbalmosdreamreal: I'd have to do some digging, but the 6-bit baudot thing I think still lives on today in some ham radio digital protocols
- dreamrealsbalmos: wouldn't surprise me AT ALL
- dreamrealI just didn't have a good answer for para so thought I'd look into it
- sbalmosdreamreal: when do you have a good answer for para?
- dreamrealscrew you!
- sbalmosweeeeeellllll...
- ParaMmmhh.
- ParaI mean I'm kinda not surprised, as I've been trained to use this piece of tech https://en.wikipedia.org/wiki/Sanomalaite_M/90 (it's basically SMS/pager in brick form with self-corrupting memory) and military used to be the source of things which became standards all over.
- nevetSanomalaite M/90 - Wikipedia
- javabotPara's title: "Sanomalaite M/90 - Wikipedia"
- sbalmosmy memory self-corrupts in much easier fashions
- ParaNobody seemed to know if it was a feature or a flaw.
- Para"Oh don't worry, it'll just give you something to do while on night watch.", something being re-entering the encryption keys and doing connection tests :P
- ParaNow that I think of it, maybe the reason we then came up with all the mobile tech is basically going "fuck this" nationwide on those things.
- jreicherPara: if you define Stream product(Array x, Stream y) then you can do n-way easily by iteratively calling that.