freenode/#lisp - IRC Chatlog
Search
23:03:19
aeth
I wonder the implications of writing something like that in Common Lisp. e.g. Memory usage is higher, so I guess you'd want to put it all in one process instead of having dozens of daemons (a few hundred MB vs. GBs of RAM usage)
23:04:43
aeth
(What I mean is that the fixed cost k in RAM usage is much higher for each process, so you could save 19k worth of RAM of you use one daemon instead of 20)
6:22:29
flip214
aeth: well, if you can share all the library and executable pages across all the processes, and need only 1MB of dynamic space or so, actual RAM usage might be the same or even smaller (because you don't need all the shared libraries' jump tables)
7:43:14
p_l
aeth: I've been looking at implementing an alternative, using ECL, though probably statically linked
7:44:43
jackdaniel
I recall someone has created init binary with ecl which was directly called by linux
7:46:52
p_l
there's nothing special about init binary except that PID1 should never exit (and of course it better reap processes)
7:52:12
p_l
also, there's no need for PID1 to use anything that requires dynamic linking, and there's no way to atomically replace multiple files, so with glibc's insanity you can get a moment where it reloads while the dynamic libs are incompatible with each other
7:52:44
p_l
yes, a freak case, but supposedly so is getting into race condition between systemd and your fingers
7:54:27
jackdaniel
since it may be bundled with the kernel it often plays crucial role when you are developing embedded boards
8:02:05
ralt
the discussion reminded me of this old project as well... https://gitlab.com/ralt/linit/
8:04:03
ralt
didn't go anywhere, but kinda shows where the limits of some implementations can be for being so high-level. At least for the old me.
8:04:57
p_l
ralt: might have weird interactions between what shared libs were present in "normal" environment and later
8:09:17
minion
The URL https://gitlab.common-lisp.net/users/sign_in?secret=68dbbefd will be valid until 08:15 UTC.
8:14:12
ralt
speaking of, Shinmera I was thinking if it was fine with you to fork portacle to add some custom startup on top of it. I'm thinking of writing some software that would essentially give you an API in slime, so having a bundle+custom startup would be perfect for me.
8:16:09
sam47
I was trying to register account on https://gitlab.common-lisp.net. It forces 2FA. I have no smartphone, so, I can't have account.
8:19:15
no-defun-allowed
It can be any other OTP generator, and I believe there's desktop generators.
8:19:36
no-defun-allowed
I was pissed off initially that the site suggested it had to be Google Authenticator, but any will do.
8:20:03
Duuqnd
AFAIK the most used type of 2FA is an open protocol, so there should be desktop programs
8:20:08
minion
The URL https://gitlab.common-lisp.net/users/sign_in?secret=cde8f4b9 will be valid until 08:30 UTC.
8:20:23
jackdaniel
me too, but I've suggested a change to easye and we he has escalated it to gitlab.com project and now it suggests also freeotp (and f-droid repository) as an alternative
8:20:46
no-defun-allowed
However it seems it breaks every so often, since FreeOTP isn't giving me valid numbers any more.
8:21:12
minion
The URL https://gitlab.common-lisp.net/users/sign_in?secret=cde8f4b9 will be valid until 08:30 UTC.
8:21:25
no-defun-allowed
This is the second time it's happened, so I think something I've done is wrong.
8:22:25
Duuqnd
antoszka: Authy has given me a bit of trouble in the past. Specifically, Twitch requires authy and didn't let me use my normal OTP program.
8:23:38
p_l
google authenticator tends to be mentioned because it's easier to tell people "Google Authenticator compatible" than have them remember the right OTP variation
8:24:37
no-defun-allowed
Huh, I set my phone to use the network provided time and it works now. Yay.
8:25:07
ralt
who am I supposed to contact if trying to register on gitlab.ci.net tells me "Your account has been blocked"?