a-dub

40 ms·

691.
by a-dub 4y ago
i thought sha-2 included the message length in the head padding to prevent length extension attacks? just read the link you provided: it's sha-224 and above. i wonder if a similar issue exists in the sha-224+ padding code.
692.
by a-dub 4y ago
when i was a kid one of my big ideas was to create a wattage normalized time/power coding scheme for microwave oven cook programs and to then equip them with barcode scanners for automatic programming based on codes that manufacturers
693.
by a-dub 4y ago
i always thought that elasticsearch would make a good host for a ML enabled datastore. building indices and searching them are similar computational paradigms to training and inference and the scaling framework would lend itself well to bot
694.
by a-dub 4y ago
that's fine. halted security updates render the device unusable. halted feature updates mean you got what you paid for.
695.
by a-dub 4y ago
this is correct. there ought to be a law where if hardware is bound to software, and the software goes out of support in a way that renders the hardware useless, then it must be open sourced. notably post qualcomm pixel phones from google h
696.
by a-dub 4y ago
it went away because it was resource expensive. even with copy-on-write semantics, those backend httpds with an entire perl runtime in each core were pretty huge and inefficient. the models that replaced it were single process multiple thre
697.
by a-dub 4y ago
> 2. Sounds like the expense means it will only be installed in offices, not people's homes. Which, again, is still useful as long as it hits the 3rd bullet point. honestly, with all the cool stuff coming out of computer vision thes
698.
by a-dub 4y ago
i love how all the people in the demo videos are excessively and unnaturally expressive in order to show off how it captures depth. would be neat to try, would be curious how much the effect holds up over time (vs flat video conferencing) a
699.
by a-dub 4y ago
many serious applications back then built by developers who cared about performance and latency made use of mod_perl. it put the entire perl interpreter (which then loaded perl modules that defined the application) into the preforked apache
700.
by a-dub 4y ago
everything old is new again. this was the concurrency model for pre 2.0 apache. (i think in 2.0 you could pick)
701.
by a-dub 4y ago
if you find that you are asking a lot of questions yourself and you are finding the answers to be unsatisfactory, maybe you should consider hiring a consultant to help you.
702.
by a-dub 4y ago
i just use forward and backward regular expressions to jump around. it's fast, flexible and basically second nature.
703.
by a-dub 4y ago
right, but massive scale production use in the google crawler to index the entire internet when that was at the bleeding edge was state of the art before the art was even really recognized as an art. i don't even think they called it A
704.
by a-dub 4y ago
when combined with minhashing it approximates jaccard similarity, so it seems it would be bounded by that.
705.
by a-dub 4y ago
10? no, it's more like 20+. lsh was a core piece of the google crawler. it was used for high performance fuzzy deduplication. see ullman's text: mining massive datasets. it's free on the web.
706.
by a-dub 4y ago
i think some logic design courses are like this. you're given a fpga dev board (that has a bunch of peripherals and ports on it) and you code up a cpu and a few memory/io controllers and then in order to do something you have to a
707.
by a-dub 4y ago
if you're in gf2 on a traditional cpu and care about wall clock time, i start to wonder how something like simd xors and parity would compare...
708.
by a-dub 4y ago
i'm guessing most of that stuff is register allocation and cache management. rl-searches could potentially rewrite entire algorithms. edit: (more musing) maybe the future of software engineering will be authoring bullet proof acceptanc
709.
by a-dub 4y ago
this is cool. i suppose it's only a matter of time until we see optimizing compilers that use transformer-rl style searches for subsets of their optimization and codegen.
710.
by a-dub 4y ago
a good computer organization/design or intro computer architecture course will have the same effect. the classic text is computer organization and design by patterson and hennessy.
711.
by a-dub 4y ago
i don't know how chess ratings work, but i suspect it could be done with a secured and monitored computer in a secure facility where some small number of games could produce a rating with decent confidence interval on it?
712.
by a-dub 4y ago
as a rule it is decidedly uncouth in our society to use population statistics when speaking in terms of ability or potential, as the act of making those statements can bias real world abilities and potentials by setting expectations of limi
713.
by a-dub 4y ago
doing the statistical analysis and all is pretty cool, but honestly, can't they just challenge him to play at his rating in a secure environment? his learning curve is a little steeper than the others... that doesn't convince me t
714.
by a-dub 4y ago
yes, this has been my joke. "maybe it won't be a recession. maybe it will be. what i can tell you for sure is that it will be the first of its kind in real-time, full surround, three dimensional, quintophonic social media stereo.&
715.
by a-dub 4y ago
i remember tymnet! their local dial-up pops fueled a lot of the commercial information services (compuserve, genie, prodigy, aol) that were popular in the 80s and 90s.
716.
by a-dub 4y ago
ahh yes, the exciting field of anus computer interfaces!
717.
by a-dub 4y ago
or better yet... follow the wikipedia model and have a parallel cheater chess league where there are cash prizes for proving after the fact that wins were the most sophisticated and undetectable cheats.
718.
by a-dub 4y ago
yes but your ability to detect rf noise from electronics, even well shielded electronics, is greatly enhanced in that environment.
719.
by a-dub 4y ago
they should play a game inside one of these: https://www.ets-lindgren.com/products/shielding/rf-shielded-...
720.
by a-dub 4y ago
and a key point regarding the two: contract/freelance rates are higher because there's no commitment from the employer regarding a stable stream of future work. also they're short term, so the actual gross difference is typic

More ›