900+ entries in 0.023s
phf:
http://btcbase.org/log/2017-12-27#1759185 << i sometimes wonder if there's some clever way to link back split posts (i mean in cases where urls are lost, perhaps by looking at cases where 250 byte message, followed by same person within ~~<3s or somesuch)
☝︎☟︎ phf:
http://btcbase.org/log/2017-12-22#1756823 << abbrev has a funny history: "заключенный красноармеец" because of early soviet labour army experiments, later renamed to "заключенный каналоармеец", i.e. builder of беломорканал, to make it more politically correct. initially the abbrev had implications of forced labour specifically, but only later was generally applied to all prisoners. (probably bec
☝︎☟︎ phf:
http://btcbase.org/log/2017-12-22#1756696 sort of reminds me of this guy at burning man, he was a chubby dude, running around completely naked, probably rolling hard, telling everyone how "oh my god, i can do anything i want!!1". the average reaction was, that's good, honey, but be a dear and keep your ballsack away from my face, there you go
☝︎☟︎ phf:
http://btcbase.org/log/2017-12-20#1755450 << i've picked up a book in new orleans, “Symbolic computation : applications to scientific computing”, it's all macsyma and various sussman style scheme hacks for exploring dynamic systems, but an interesting trick that almost all the code uses, which i guess was sop in lisp world at some point, is to do visualization and control on a lisp machine, but do the heavy numeric simulations in fortran 77. the in
☝︎☟︎ phf: it takes about 12 seconds to walk the log for all the messages containing "
http" and then narrow it down with a somewhat hairy regex that i found somewhere some time ago
phf:
http://btcbase.org/log/2017-12-04#1746081 << i can, by the way, give you (or anyone who wants it really) an ongoing feed of urls, which might be a bit saner, than trying to wget a bunch of html pages, and then attempt to reverse them
☝︎ phf:
http://btcbase.org/log/2017-12-03#1745764 << i recall we ran into the issue of vpatch-ing binaries that hasn't been resolved. wp admin interface relies on a lot of png's and gif's that it's not clear how to stuff into a vpatch without a superfluous "make" phase
☝︎☟︎ phf: ok, here's a more integrated PoC
http://btcbase.org/log-search?q=from%3Amircea+god the image is being generated full scale, but then force by the browser to be 1000px wide. (fwiw, while still cheap, it is more expensive to resize it, than to spit out the whole thing)
phf: ok, i have some idea why
http://btcbase.org/log/2017-11-23#1742652 because i've done it in the past. when you're debugging issues for complete newbs, you basically instinctually looking for two things, imbalanced parenthesis and misuse of intrinsics. yeah you could copy paste that stuff to your emacs buffer, but if what you're doing is essentially social service work, you get burned out on doing deep help pretty fast, hence you want to keep it in the p
☝︎ phf: we can of course specify our own subset of sexp that we use for interchange (along the lines of what
http://www.islisp.info/ did) but that comes with all the "specify our own interchange format" caveats
☟︎ phf:
http://btcbase.org/log/2017-11-20#1740959 << i think i said that in logs before, but sexp is a poor data exchange format. read operation relies on current state of *readtable* and traditionally readtable is customized to specific operator needs. there's a default *readtable*, but it's too unconstrained to be used for deserialization of untrusted data, since it, for example, supports arbitrary code execution, circular structures, etc.
☝︎ phf:
http://btcbase.org/log/2017-11-20#1741179 << you can just ignore the whole "string" question in first version, McCarthy's lisp used symbols instead of strings (that's why early nlp code, like eliza all come out as DOG SAID, HELLO) and the only operation you could do at some point was read and eq.
☝︎☟︎☟︎ phf:
http://btcbase.org/log/2017-11-20#1741280 << it doesn't, logotron necessarily has to interpret the nature of bytes (because the display layer is explicitly encoded). so the log collection and storage is 8 bit clean, but display makes assumpetions for rendering purpose. specifically i assume that log is utf8, but if utf8 decoder fails on line, i fall back to latin-1 (this is a traditional irc mechanism, details of which have been discussed)
☝︎ phf:
http://btcbase.org/log/2017-11-14#1737529 << that doesn't sound right, read and eval are distinct phases, by the time you get to eval you shouldn't be operating with strings when but instead with interned symbols (i.e. things that can be eq'd in lisp and pointer equivalent on c machine level)
☝︎☟︎☟︎ phf:
http://btcbase.org/log/2017-11-13#1737247 << it looks like a properly structured scheme evaluator, but it's ~explicitly~ lacking a native cons, which might be a very good exercise for whoever™ adding a static allocation space, adding mark-and-sweep, then all those To_Unbounded_String look like they can be simply search/replaced
☝︎☟︎ phf: asciilifeform: that's josh dersch, he's a seattle retro computing enthusiast. specifically he works on hardware restoration (including a symbolics) for paul alan's
http://www.livingcomputers.org, i've met him once or twice