NeXeNi love backported stories, they get all the details right, ...or wait do they get them wrong? hell i forget how the story goes
DrixtanI am building native binaries with java projects - In Your Teeth Mr. GoLang... No but seriously, no trash talking here; that's pretty awesome if you ask me - but since you didn't... :D
javabotdreamreal's title: "Googlebook: Designed for Gemini Intelligence | Coming Fall 2026 - Googlebook"
nevetGooglebook: Designed for Gemini Intelligence | Coming fall 2026
dreamreal"phone with android 17 or above required" and what a freaking name
Drixtanis it free? since we are the product, I guess it's going to be free isn't it?
dreamrealsomehow I doubt it
Drixtanafter the mac neo, I don't know what google expect with this one.
Chronosdreamreal: Impressive lack of details on that Googlebook page...
dreamrealyeah, I laughed
dmlloydhere's a question for any windows dev people here: why can I not resolve the `GetComputerNameW` symbol using FFM? do I have the wrong symbol name, or do I need a specific library for this, or does the JVM block some symbols, or...?
ChronosI could be wrong, but it looks like it's Google's first Android-for-the-desktop-that-replaces-ChromeOS.
GreenResponseit only lacks Musk’s NeuraLink to ask Gemini just by thinking–save your vocal cords for dealing with other humans!
dreamrealneuralink would ask grok anyway
GreenResponsetrue
deebowhere's my AskJeevesBook, with a mos6510
dreamreal8510 now
dreamrealActually, I can't remember the differences between the 6510 and the 85xx series the commodore 128 had - page switching for RAM banks, at least, I think?
* dreamreal runs to ask whatever the AI du jour is
dreamrealoh, 8502
dmlloydthe differences were very very minor IIRC
dreamrealyeah; 6510 added some I/O states, 8502 was the memory flipping stuff
dreamrealyou RC
dmlloydin general the customizations to the 6502 that were added *after* the initial runs were all really minor
dmlloyduntil the 65C02
dmlloydwhich was a redesign
cheeseri still want to write an emulator for that cpu
dmlloydit's a nifty design, very efficient and clever
dreamrealdmlloyd: incidentally I used your methodhandle article for work yesterday - the semantics around invokeExact() are ... interesting. I ended up going with invoke() instead, but I literally could not call invokeExact() on a method that returned a long without consuming the long.
dreamrealI'm going to have to defend the way I used it ("couldn't this have been a lambda?") but I'm sticking to it
dmlloydyeah you have to consume it or else throw it away ahead of time. `mh.invoke()` is semantically identical to `mh.asType(<the way you are going to invoke it>).invokeExact()`
cheeserstupid things I want to write: 6502/C-64 emulator (using javafx, natch), ext4 fs, mongodb/lsm in-memory (h2-inspired mini version)
cheeseri once debated writing a minecraft mod layer on top of mongodb query visualizations
dreamrealdmlloyd: I don't even know how that's *enforced*. invoke() was fine, and I almost went with the lambda but screw that
dreamrealI was mostly vaguely amused that the method I was invoking *returned* a long... and did nothing with it. A *private* method!
dmlloydwell `invoke` is a special magic case
dmlloydimagine that it's like a macro, and the JVM expands it on first use
dreamrealA smart developer would say "why not make those void" but um
dmlloydwell you can make it void: `mh = mh.asType(void.class, arg1Type, arg2Type, ...)` boom now it's void
dmlloydand you can `invokeExact` it
dreamrealHah. Is that worth it, though?
dmlloydif you call it more than once, yes
dmlloydthe JVM can theoretically inline any `invokeExact` on a method handle living someplace stable (like a final static field) into basically being equivalent to calling the method directly (possibly including more inlining)
dreamrealhmm, hmm. I'll have to see how often that gets called - it's mostly called in an outlier data condition on multipart emails. And I HATE that method - it's nearly 200 lines long, and the cyclomatic complexity is unbelievable.
dreamrealFor those who have been around #java on IRC for a long time: MaggieL passed away on May 7. Hasn't been seen on libera that I know of, and left freenode a long time ago, but MaggieL was knowledgable, irascible, and a valued and trusted friend and contributor. Pour one out for her.
jbosmansdidn't know her but may she rest in peace + cheers with a belgian beer
dreamreal"The sunk cost fallacy feels like conviction if you squint hard enough. Look, we get it, Groovy sounded like a great idea at first."
dreamreal... one part of that is going to get dropped in the next draft :D
ParaGroovy was fun.
dreamrealSo was UML.
ParaNah.
dreamrealpfffft, now we know yet another person who never used together/j
ParaI had to learn Rational Rose to get through school.
dreamrealI thought together/j was good enough that I was willing to try IDEA
cheeseri miss the remote collaboration it had. too bad the internet wasn't fast enough to make that practical then.
Paratrac was magical
ParaWe've lost so much.
dreamrealAt least we have copilot 365 teams with copilot to pilot us co-ly along with a copilot (tm)