700+ entries in 0.019s
sina: bang it out, see if it sucks or not
sina: mircea_popescu: re pseudocode is exactly how I view it
sina: asciilifeform: feel free to check out my github :P
sina: most "something elses" don't meet my "seemed interesting" criteria
sina: saw spec on trilema, seemed interesting, why not
sina: yeah but I am not making this software and saying you sould use it for serious, or even any, thing
sina: btw asciilifeform thanks for checking out guix, I thought you might be interested after reading old logs of you complaining re ripping dbus out of emacs
sina: yes and it's understood, I'm not trying to make "gossipd for alfs dystopian future" just have some fun for me
☟︎ sina: ok, thanks for validating mircea_popescu, tearing down the shithost
sina: as it currently sits with deedbot, I currently don't, but can, run my irc connection on compartmented box away from gpg
sina: in irctron, you can login to a remote node and keep your privkey operations local
sina: although it could be decoupled somewhat
sina: in the gossipthing I made, the transport layer is tied to the encryption layer
sina: asciilifeform: I do like that idea
sina: and once all components are happy, can be rewrote in a not shitlang
sina: asciilifeform: it is designed to easily rip/replace the RSA and other componenty bits easily
sina: if you like it too, that makes me happy
sina: but again, disclaimer, am not doing this to infiltrate tmsr with shitlang, as asciilifeform seems concerned, just for my own fun!
sina: anyway, if youre happy with that generally as a "first draft" I am keen to refine incrementally
sina: as you can see in the gossipc stdout, it's all RSA, stateless whatever style
sina: mircea_popescu: ssh root@45.77.66.53 in two windows, one 'gossipd'. other 'gossipc --send-message --source whoever --message whatever' and then 'gossipc -g' to view msgs
sina: yeah cool that's pretty much what I thought after yoru comments re sqlite/rsa being the main things needing changifying
sina: 'ma tear down the shithost
sina: mircea_popescu: you wanna play with this gossipthing yet?! otherwise I'
sina: literally told me you don't believe in time!
sina: y'all don't have nothin'
sina: best twitter quote I saw today: "US is free as in mattress"
sina: I am out. hope all have a wonderful day
sina: can leave me a note with anything you require
sina: ben_vulpes: I am about to head out the door so currently not the best time, but if it would help in anyway I can donate some compute to the effort
sina: ben_vulpes: will you include pypy?
sina: well, I guess lets see how the benchmarks play out over a larger dataset, maybe it evens out over a certain bitlength or bytesizer
sina: cos that'd be pretty interesting
sina: ben_vulpes: did you use sbcl the other day when you mentioned golang impl was faster than lisp impl?
sina: it does sound like lisp is doing something JIT-like, except you can "re-JIT" at any time during execution?
sina: anyway, it does sound like `time` should be fine
sina: black box doesn't have to be invoking from CLI each time tho, you could write a daemon around each impl and measure how long it takes to return a value
sina: or you are OK with a blackbox, in which case can just use time and avoid calling it with small loops where the runtime startup cost dominantes `time`
sina: my point was, either you *really* care about evaluating the actual main loop, which is a fair apples/apples
sina: ben_vulpes: you asked what else "it" can depend on, where "it" == whether or not runtime startup is a cost of the program or not
sina: <+ben_vulpes> runtime startup is a cost of the program, innit? <<
sina: just as an example of "anything else?" re "depends!"
sina: ben_vulpes: sorry. I am referring to pypy JIT vs python
sina: its reading something closer to native compiled code
sina: faster in the run because its no longer "interpreting"
sina: same as LuaJIT vs Lua
sina: for example, pypy is much faster than cpython for long running programs, because it Just In Time compiles
sina: well, I dunno too much about lisp, does it "JIT" for long running programs?
sina: that might be a fairer blackbox test?
sina: if I make an mpfhf daemon, then no
sina: of the main loop iterating through M
sina: so if you really want super precise, apples <=> apples comparison, you would need to instrument performance on a per lang basis, no?
sina: because, for example, python and lisp, probably most of the time will be spent in starting the runtime/interpreter than actual computation, unless you're doing larger sized stuff
sina: here is the thing, it depends on how anally you want to measure
sina: is that all you want to measure? why not just use `time` then?
sina: e.g. "how long does it take to hash document of N bytes size in M bits hash" with varying N and M?
sina: ben_vulpes: oic. and why not just using black box testing?
sina: ben_vulpes: can we roll back and start at the usecase?
sina: mircea_popescu: around? any time to play with gossipthing?
sina: but again, I haven't used that one
sina: systemtap is basically "dtrace for linux" I guess
sina: you can run it on fbsd
sina: it's pretty awesome
sina: ben_vulpes: what is wrong with perf for your usecase?
sina: I heard good things about systemtap but never used it myself