so...
how would i start up the other core of my Eden U4200 machine from within DOS, and then make it do something useful?
also has anyone actually done anything like that before? because it could be useful for not just DOS but RISC OS too - essentially turn the cores not doing anything into custom coprocessors
i have a feeling that it's trying to resolve the IP address i typed into the proxy field as a hostname, which is so obviously a bug i don't know where to begin... or, more importantly, how to get around it
ed. well, adding the proxy server to hosts doesn't help, so i guess it's not that
apparently it wasn't working six years ago, and hasn't been fixed since...
i was hoping to see what WebPositive is like as a browser, but i can't make it use a proxy, so that's kind of that
haiku's developers might have adopted an anti-LLM stance, but... i guess not everyone got the memo:
https://github.com/ltouati/HaikuRemoteDesktop
This application was mostly vibe-coded using Antigravity and Gemini 3.0
when i program in a REPL, i am having a conversation with a machine in its language.
when you use an LLM, you are having a conversation with a machine in your language.
we are not the same.
Without AI, this project would not have been feasible for me as a hobby project.
then i'm afraid the project was not feasible for you as a hobby project. which is sad, because had you tried it without, you might have surprised yourself.
otter-browser doesn't work either. frankly, NetBSD/i386 is beginning to feel a bit like abandonware...
do i start again with NetBSD x64, or do i commit to the whole "learn a new OS" thing and plough on with Haiku? i mean, the big problem i have with Linux x64 is that it encourages the worst overcommit behaviour; maybe Haiku will be better than NetBSD in that regard? (unless it's now just a case of "well Firefox and Chrome expect it and won't start without it, so we have to")
one could argue that i'm compromising myself here, because i'm, let's say, not the most detail-oriented person around, and am prone to making the kind of errors that sttic typing catches. and yet, still i can't abide it...
am i alone in having a visceral dislike for statically typed languages?
i don't tend to argue much about this, because (a) i get shouted at by just about everyone and i can't bear that¹, and (b) i'm not even sure i have arguments about it, except "if you're going to frontload computation anyway, you might as well fully embrace compile-time execution". it's really just, as i say, visceral. i think it goes along with my distaste for bureaucracy in all walks of life
actually, that last one - the adding a note on someone's account - is the one that really pisses me off. when it was introduced, it was a really easy way to make a quick note of why you followed someone, or why you blocked them, or why you were thinking about blocking them, or... anything else really. it was so easy to use that it didn't break flow at all.
i guess someone decided it was too easy to use. so they made it a modal dialogue that could only be accessed from a pulldown menu instead. and instead of the notes field always being there and inviting, it now doesn't appear until you've made a note.
all i can think of is that someone who never even used that facility once decided to "improve" it without asking anyone who might have used it a lot, because those people's use cases didn't fit into their model of what profiles were for, or should look like
and that way of developing stuff is bullshit and has no place in something like #mastodon which is supposed to be all about community.
the sooner mastodon can end its love affair with modal dialogs the better
i don't want to have to stop everything just to change the privacy settings on my post
i don't want to have to stop everything just to add a note on someone's account
these are things i could do before, more importantly, without stopping everything. but now they\ve become modal dialogs. this is such a retrograde step.
ok, so why does enabling CSM support on the Purported Netbook disable the built-in keyboard?
(it's a Geobook1m; the UEFI is an AMI one with all the trimmings... and apparently at least some of the bugs)
on the bright side, it does boot Haiku (and even offers a 1920x1200 display mode - i don't know whether that's scaled or the inherent LCD resolution), and seems to run it quite well... albeit, as usual, with no touchpad. i'm really very tempted to just stick Haiku on it and go!
so anyway, the upshot of that is that i'm now very much in the market for a more sensible editor with Wordstar keystrokes. it's that, or go back to using David Nye's little editor under dosbox or something, and accepting that lines should only ever be 80 characters long (and i should only ever see 24 of them at once)
ffs....
so. using joe in jstar mode, i just wrote out a section of a file to another one using Ctrl-K W. or at least, i thought i did; i must have pressed Ctrl-K S instead, so it ended up renaming the file,
in the meantime, i went to edit that file, and renamed it
so when i came to save the original file, which was otherwise unsaved, i typed Ctrl-K X, which should have saved all changes and exited, and... joe said
File not saved as no unsaved changes since last save
it didn't even do the most cursory check: whether the damn file is even still there! that's just... infuriating. so that's basically an afternoon's notes lost.
fuck.
why this is weird:
well, this is the sum total of the contents of fcopy.4th:
0 VALUE idfR
0 VALUE idfW
\ - FILE-COPY-BUF VARIABLE FILE-COPY-BUF $80000 ALLOT
0 VALUE FILE-COPY-BUF
$8000 VALUE FCOPY-BUF-SIZE
: COPY_FILE ( adrR lenR adrW lenW -- )
( \ idfR idfW -- )
FCOPY-BUF-SIZE ALLOCATE THROW TO FILE-COPY-BUF
R/W CREATE-FILE THROW TO idfW
R/O OPEN-FILE THROW TO idfR
BEGIN
FILE-COPY-BUF
FILE-COPY-BUF FCOPY-BUF-SIZE idfR READ-FILE THROW
TUCK idfW WRITE-FILE THROW
FCOPY-BUF-SIZE <>
UNTIL
idfR CLOSE-FILE THROW
idfW CLOSE-FILE THROW
FILE-COPY-BUF FREE DROP
;
: FCOPY ( -- )
PARSE-NAME PARSE-NAME COPY_FILE ;
nothing in there looks even slightly controversial
ok. through n-furcating .()s, i've narrowed it down to the line in autoexec.4th that says
REQUIRE FCOPY ForthLib\tools\fcopy.4th
for whatever reason it doesn't come back from that.
there's also https://github.com/GRAVITYDIV10/uefi-forth , which is an ITC implementation, but i don't know if that means "more conventional"
having thought "someone must have made a UEFI Forth by now", i went looking for one. sure enough, there are several, but this one might be the most complete: https://github.com/mak4444/fasm-efi64-forth/ it's native compiled, and i doubt it hews particularly closely to anything standard
it didn't work on the newly UEFI'd Chromebook, of course - is that a bug, though, or just a reflection of coreboot not actually being a completely generic UEFI but only as much as is required to boot an OS? [ed, apparently not - see downthread]
but it came up just perfectly on the ASUS i3-6100. which suggests either the above, or it's been tested mainly on an ASUS UEFI
but i have plenty of UEFI devices - i shall report back in time...!
ok: