sylware

43 ms·

901.
by sylware 2y ago
How do I test that?? I code in vim (or nano) with some basic sh scripts (certainly not bash). Then I would remotely query gemini via curl, or wget, something along those lines. I could not see any pointers on some "web API" docume
902.
by sylware 2y ago
This is why I did speak up: "softbank" hides "ARM ISA" from the people unaware of this.
903.
by sylware 2y ago
Is there some truth in the following relationship: o1 -> openai -> microsoft -> github for "training data" ?
904.
by sylware 2y ago
ARM isa instead of RISC-V, bad.
905.
by sylware 2y ago
kewl! Does a vulkan3D backend make any sense at all for this engine?
906.
by sylware 2y ago
restore noscript/basic (x)html interop on all critical/dominant sites where it did a good enough job already years ago?
907.
by sylware 2y ago
How to say this... it was not in the first place. And it is not specific to the US, it is the external cloud operator which is the issue. It is a very complex matter. Roughly speaking, if you rely a lot on information systems, in the end yo
908.
by sylware 2y ago
Yes, it is since you can with a little C preprocessor abstraction assemble with fasm/gas/nasm.
909.
by sylware 2y ago
With that type of news, who does not think "Musk's suicide squad". :P
910.
by sylware 2y ago
Solution: restore noscript/basic (x)html interop, where it can do a good enough job (like most online services where doing a few years ago...).
911.
by sylware 2y ago
Have a look an their code, it is obvious. Often you have to figure out what actually the macros does, and I remember it was not that straight forward. And the macro language is specific to nasm. What to do: unroll the macros and/or use
912.
by sylware 2y ago
A gigantic mistake was done in much of ffmpeg assembly: They are abusing nasm macro-preprocessor up to obscene levels...
913.
by sylware 2y ago
I use links2 and lynx web browsers. The issue in EU is Big Tech has many "minions" in public administrations and local critical online services (many dominant online services are theirs anyway). They manage to make hard dependent
914.
by sylware 2y ago
But musk is doing the right thing: he is literaly destroying the value of words, probably on purpose. I guess he wants you to just look at what he actually _does_ (after _really_ verifying he _really_ did that...). This is far fetched, but
915.
by sylware 2y ago
dwm and st, everyday.
916.
by sylware 2y ago
... and it is getting 70% of gcc speed...
917.
by sylware 2y ago
Good point. But you only need another one reason to make c++ a definitive nono (and similar languages with ultra complex syntax, like rust, java, etc): developing a _real life_ c++ compiler alternative (citizen/state-sponsord/smal
918.
by sylware 2y ago
AI????
919.
by sylware 2y ago
I thought it was using libfirm... not that c++ diarrhea of llvm... :(
920.
by sylware 2y ago
Oh... What I read did not mention that many backends. Don't read the doc, read the code... :( Definitely worth some benchmarks, until libfirm SDK is reasonable: there is a bad start with cmake (c++) and I have no idea if it requires ex
921.
by sylware 2y ago
You mean Circle can output correct intermediate code for the latest gcc code? Or valve mesa aco code?
922.
by sylware 2y ago
Missing RISC-V code generation. But it is written in plain and simple C99, so it is at least much less toxic than LLVM. I wonder how cparser (did not check if it was plain and simple C)+libfirm compare in performance to cproc+QBE on my ben
923.
by sylware 2y ago
wrong, point missed by miles away. Maybe, "normal"/"average" devs would help you to get the point.
924.
by sylware 2y ago
Do you remember the time you could buy on amazon.com with lynx? (or links2). And all the other dominant online services which broke noscript/basic (x)html to the benefit of the absurdely and grotesquely massive and complex (including t
925.
by sylware 2y ago
On the same way than cproc+QBE I guess, 70% of gcc speed in my benchmarks. The more real-life alternatives to those abominations of gcc and clang, the merrier. I wish the linux kernel devs did care to keep the door _reasonably_ open for suc
926.
by sylware 2y ago
Is the next step ML-inference fusion? aka artificial small brain?
927.
by sylware 2y ago
... and valve is unable to generate clean 64bits ELF binaries, like the "below average game developer" for the blink gogol web engine they use as the steam client ("cef") since it requires pressure-vessel.
928.
by sylware 2y ago
What's very nice with wayland protocol, clients have to query the availability of this sub-protocol and gracefully fallback to alternative code paths if not around.
929.
by sylware 2y ago
DeepSeek is good enough, and _REALLY_ open, unlike the lie of "openAI". Won't last since more trained and bigger large language model and "reasoning" language model will arrive. And we know how Big Tech does work: t
930.
by sylware 2y ago
AI?

More ›