i jumped off the map, got respawned twice, our fuse was afk since 7min in, and i punched the last guy to death because we both ran out of ammo. psycho game. 10/10
hit again by less scrolling perf decreasing linearly with current-line-length-seen; found https://bugs.debian.org/36318 (five digits!):
Subject: less: [PATCH] backward scrolling in binary files is very slow
Date: 19 Apr 1999
> I barely remember writing the code at all now,
Replying to @nabijaczleweli
oh, the forwarding address is bug-less@gnu.org; that explains it
i just wrote "juicy semantics" and "oozing with NUL-related issues" i think im hungry (and V4 tr is ass but what's new)
Replying to @nabijaczleweli
in what is undoubtedly the /worst/ mailing list interaction ive ever had (plus the worst archive webshit in the game: cf. top left); ive finally managed to post to the god-forsaken uclibc-ng list: https://mailman.openadk.org/mailman3/hyperkitty/list/devel@uclibc-ng.org/thread/HOD6LXGDGUTVFJIVJ6RXCDUC75NGXJRX/
Replying to @nabijaczleweli
(where "worst" is defined as there's an account system, in addition to normal subscriptions, even if you subscribe you cannot quiet-subscribe, even via the mail commands, and 250 OKed and queued my patch before subscribing, but dropped it(?), and doesn't say it's subscriber-only)
Replying to @nabijaczleweli
not an hour since i subscribed and i already got spam shat out to the list address lmao
Replying to @nabijaczleweli
> X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; emergency; loop; banned-address; member-moderation; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header
SO DON'T FUCKING FORWARD IT
found a mandoc bug while reducing a manual for a mandoc bug report. great one today.
Replying to @nabijaczleweli
Intended: https://inbox.vuxu.org/mandoc-discuss/20220605163009.6q2rxkbky6bxcx3t@tarta.nabijaczleweli.xyz/T/
Bonus: https://inbox.vuxu.org/mandoc-discuss/20220605161635.chte3k2gotoceerf@tarta.nabijaczleweli.xyz/T/
Replying to @nabijaczleweli
the first appears to be a long-standing and difficult-to-fix groffism, the latter got griefed by an unhinged czech who hosts the PDF covid certs for his entire family on his mail domain (and is /very/ lucky to be out of reach)
Replying to @nabijaczleweli
ruh roh
<- this user consumes deadly amounts of arsenic from imported american rice
Replying to @nabijaczleweli
:/
Replying to @nabijaczleweli
lead in your paint, arsenic in your rice, my brethren in grass heavy metals dont even get you high
бублики moment :0
Replying to @nabijaczleweli
(way too big, the portioning didn't work out right because i had either really big eggs or exceptionally wet flour; still fucks doe)
Replying to @nabijaczleweli
i was told i fucked the white balance and these looked like they were potato, when they are, in fact, regular grass. please consider this correction
another day, another coreutils moment in tr land: https://bugs.debian.org/1012447
tr is the worst utility in UNIX history. no-one has /ever/ implemented the blue parts. non-4.3BSD-Reno tr (i.e. the one this "specifies") stops processing string[12] when it gets \0 and strips NULs from the input, which means it's not "Eight Bit Transparent"
Replying to @nabijaczleweli
"8-bit transparency" is unexplained. 2.2.1 is literally a list of all the utilities that are "noted to be 8-bit transparent". goddamned XPG3 moment.
Replying to @nabijaczleweli
IN is internationalisation (defined as optional), UN is /unimplementable/, which is a great look for a standard.
Replying to @nabijaczleweli
wrote five pages again
Replying to @nabijaczleweli
i need a normal hobby, like drinking
Replying to @nabijaczleweli
normal day in a utility that hasn't meaningfully changed all that much since like V7
Replying to @nabijaczleweli
alas: haha preprint
Replying to @nabijaczleweli
:0 salad lookin ass
lmao, FreeBSD expand.1 -t heading is broken, because it got fixed to avoid confusion with the macro that makes the rendering be correct? https://github.com/freebsd/freebsd-src/commit/042e4ab35583f777803c194d466189090469c274
Replying to @nabijaczleweli
freebsd moment https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=264556
Replying to @nabijaczleweli
forgot i had these posts linked under unexpand in the README as "the worst time of my life"; https://twitter.com/nabijaczleweli/status/1534635723843899397
Replying to @nabijaczleweli
also: just found the coreutils bug that was biting me :) shock of shocks, it's a conformance bug. https://bugs.debian.org/1012545
Replying to @__phantomderp
you're off-shot in a tall chair with a notepad
Replying to @__phantomderp
(i can only extrapolate that this is "prior art" and falls under "3rd-party implementation effort")
Replying to @__phantomderp
well they're exercised in the wild so id say ex def ya
Replying to @eatijr
this (and a usable target/doc/) is why i use a unified target directory, which i can only recommend. ill get back to you when it calculates how big it is
Replying to @eatijr
a mere 207G on my desktop (only 2min, used to be 2h, i forgot morbing it to a rectangle (+ new cargo doc exploded on old docs so i purged like 4-6 gigs of t/doc) and 20G (3:27, 2.7x the rootfs lol) on flat computer
/that/ sounds like magic, I just do ln -s ../Rust-target target
Replying to @nabijaczleweli
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=264559 (as usual: this is worse; printf 'a b' | unexpand -t2,3 | expand -t2,3 is yields one space, not two)
Replying to @nabijaczleweli
hm
Replying to @nabijaczleweli
"we saw you from across the room and we fucking hate your vibe" type beat
Replying to @nabijaczleweli
oh great, coreutils unexpand also folds <blank>s into existing <tab>s(?) https://bugs.debian.org/1012545#10
Replying to @nabijaczleweli
freebsd moment: printf ' a' | unexpand -t4,55,68,78 replaces the first <blank> with a <tab>. it's not even close. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=264559#c2
Replying to @nabijaczleweli
okay, nix that last part for coreutils, POSIX very helpfully mixes notation; <blank> means iswblank(), but -t also says "<space>-to- <tab>" which is fun
Replying to @nabijaczleweli
oh great it does that appx. 50/50. not changing until i get an interpretation.
Replying to @nabijaczleweli
christ https://www.mail-archive.com/austin-group-l@opengroup.org/msg09779.html
Replying to @nabijaczleweli
lmfao it ate the <space> and <blank>. guess why :)
Replying to @nabijaczleweli
haha https://www.mail-archive.com/gossip@mail-archive.com/msg01613.html
Replying to @nabijaczleweli
:/
Replying to @nabijaczleweli
got an interpretation: https://www.mail-archive.com/austin-group-l@opengroup.org/msg09780.html: folding together <space><tab> is fine; "the standard is not clear" on whether you can fold any <blank>
Replying to @nabijaczleweli
true blueblik classic
Replying to @nabijaczleweli
that don't work, do it
Replying to @nabijaczleweli
pee print
is this excessive https://bugs.debian.org/1012709
Replying to @nabijaczleweli
these mfs do be tasting somewhere between rucola and radish so minor success on that part at least
Replying to @nabijaczleweli
who decided to make dirt so fucking heavy bro
Replying to @eatijr
no way in hell, only the dutch can manage to make good/seen into a damn hieroglyph
Replying to @eatijr
(OTOH, im not sure how you'd even use that to "grade homework" since you grade something by, uh, grading it 1-6, not by inscribing a symbol of power on it, and the more easteurope you go the more you sign every page on a legal document, so)
Replying to @eatijr
((OTOOH in poland you parafe stuff which is like if your signature got krulled and i have 0 clue if that's something you do literally anywhere beyond poland/czechoslovakia))
🥺
Replying to @nabijaczleweli
after decades of training and self-actualisation i became three-level table lookups and a tree traversal
Replying to @Foone
are you building for x86 or amd64? it sure looks like it should be 24 (4*4+2*2+4) on x86 to me (and 32 with 8-byte alignment (4*4+2*2+pad(4)+8) on amd64).
Replying to @nabijaczleweli
(500 of that damage during the final fight)
sensible interface moment
Replying to @nabijaczleweli
preprint, &c., &c.
Replying to @nabijaczleweli
i am hogging
Replying to @nabijaczleweli
with da flowers :0
Replying to @nabijaczleweli
(dunno if thats a good thing or whatever; simultaneously i dont really care and i dont think any daikon farmers follow me for obvious reasons)
Replying to @nabijaczleweli
the wins are like daikons to me
Replying to @nabijaczleweli
Replying to @b0rk
sourcehut
*permutes your index* sahrry sahrry *permutes your index* oh god oh jeez *permutes your index* nonononono *permutes your inde
Replying to @nabijaczleweli
all in a day's lexgrog (this is a lie, ive been writing this for like five nonconsecutive days in hour increments because it's. well its roff. its not great unless you're already shitlordified)
Replying to @nabijaczleweli
i mean, it looks cool as hell but then im a sucker for this its just Vaguely like the SysIII index
Replying to @nabijaczleweli
update: i looked, SysIII does it better; i havent improved on 1982
Replying to @nabijaczleweli
(well okay thats always been out of the question but still); also i forgot not everyone has the SysIII User's Manual on speed-dial
wha? am i undergoing a GCC moment?
Replying to @nabijaczleweli
ah, yes, this is a GCC moment, saw_nonblank and blit_rest are bools. thank.s
Replying to @eatijr
i guarantee this would 100% be a reality if that worked in slavic languages
im sorry im sorry im trying to delete it
Replying to @nabijaczleweli
-2 me https://youtu.be/s6j_PAS3788
Replying to @nabijaczleweli
finally identified the head-scratcher that rendered this so sloppily (1: note that "11 520" is shorter than "147.31"!) and renders it at /actually/ fixed width (2: note perfect alignment of all columns)
Replying to @nabijaczleweli
its kerning. of course its kerning. you remove kern pairs from the font metrics. although I already spec -n for afmtodit for C* fonts ("use with monospaced (constant-width) fonts."), this only removes ligatures, not kern pairs.
Replying to @nabijaczleweli
im so fucking stupid. at least i caught this before I posted it to debbugs
Replying to @nabijaczleweli
fixed in live
Replying to @nabijaczleweli
:0:0:0
ive finished the mass revisit so i can finally rebase my HEAD; 2k2L/53.4kB of new manual :)
sometimes for a fleeting moment i manage to forget americans build their houses from firewood + a layer of gypsum, and what a beautiful moment it is!
Replying to @nabijaczleweli
includes https://twitter.com/gray_dot_bmp/status/1521113127819034628 :) https://git.sr.ht/~nabijaczleweli/voreutils/commit/13133f12a5cc90c6976f7fa813d275c3f5e58f12
Replying to @nabijaczleweli
and an index for the HTML render :0 https://srhtcdn.githack.com/~nabijaczleweli/voreutils/blob/man/man0/index.0.html (some renders minorly broken because of a mandoc bug unfixed in trunk: https://inbox.vuxu.org/mandoc-discuss/20220606153911.w4bvcz6zqsjde7so@tarta.nabijaczleweli.xyz/T/#mb1cec7ed4e614568326572e5310d83fad3fc3bff, but hey)
Replying to @__phantomderp
what kinda [redacted]-ass CI you're using that doesn't just let you SSH in lol
currently experiencing downtime
Replying to @nabijaczleweli
blegh, truncated initrd on both kernels because it ENOSPCed, 100M /boot is too small to casually snapshot :/
ETA 18 minutes for pool import
Replying to @nabijaczleweli
so big :0
Replying to @nabijaczleweli
vindicated :0 https://bugs.debian.org/1013924
lads, pseudoscience me: what micronutrient am i not getting that makes me not fall asleep
Replying to @nabijaczleweli
(if someone says "bitches" im deactivating)
Replying to @eatijr
scarily close to not pseudoscience! thankfully, i know the moon and sun are not this close on the firmament :)
jeans are shapewear
teething again. -2
who made it summer again. we should avoid this going forward.
yo pog a bash history rewriting bug :0
Replying to @nabijaczleweli
meh, fixed in experimental :/
oh for fuck's sake i thought comm was gonna be non-problematic but no: usefully, the documented --output-delimiter=STR holds /except/ if STR is the empty string, in which case the separator is a single NUL. thank.s
Replying to @nabijaczleweli
debbugs is my single most-corresponded-with entity of recent i think https://bugs.debian.org/1014008
Replying to @nabijaczleweli
........ but not for the --total output which is separated with the empty string. what the actual fuck
Replying to @nabijaczleweli
slept 14. entering my stefania follini era?
Replying to @imhkr
i mean, from a perf stand-point the paste might need re-doing but id say thats negligible; otoh, the main differentiating factor would be having been run hot 24/7 for its service life and getting that much of electromigration, which means you'll get a shorter lifespan
Replying to @imhkr
(but thats why they're cheaper than equivalent GPUs used for fortnite (im assuming? i havent had cause to look at the New GPU Market) so you get what you pay for, so meh)
Replying to @imhkr
if you don't run your hardware until it dies then i dont realistically see how this could be an issue in the general case, yeah
what do they b lookin at
Replying to @nabijaczleweli
is this what being "in season" means
Replying to @nabijaczleweli
is there supposed to be a legume
lmfao. cyrillic уМае (that, notice, is fucked up at the top), "discordgilfcx", "AirDrop" is an Apple product, Steam branded once correctly and as STEAM twice, "Discords". anything else obvious?
post-print https://git.sr.ht/~nabijaczleweli/voreutils/commit/b34c209742ffc39dcea0b073e2f93717baa9ee58
man, I love V3. the by-initial call-out. the DIAGNOSTICS section of "nah". literal diff but self-admittedly awful. perfexion
ayo this mf flat as. how the hell
Replying to @eatijr
lmao literal heirloom UNIX brainworms but "oh wait don't you usually bitnot crc" (pictured: cksum(1) vs sum(1) compat modes)