Your twitter archive

any commutators in the chat for a pre-print?

 Fri Jul 01 02:29:24 +0000 2022

any1 coping/seething 2day?

 Sat Jul 02 14:11:40 +0000 2022

hit up the relevant netbsd/openbsd lists, they have porterboxes (well, "box" is a generous term for a fridge, but you get my point)

 Sat Jul 02 15:09:38 +0000 2022

(but also your laptop can run as many fully-featured SIMH VAXen as it has CPUs so)

 Sat Jul 02 16:10:53 +0000 2022

ive sent mail to/from 4.2BSD-on-SIMH before. Just Works. presumably if you install a system not-from-1983 you can do equally contemporary stuff like ssh

 Sat Jul 02 20:10:59 +0000 2022

https://twitter.com/nabijaczleweli/status/1524793951001137156

 Sat Jul 02 20:12:25 +0000 2022

its always "dont open up microwaves", "microwave parts are dangerous", "you can't re-use the bits anyway", but like, its very easy to repurpose the HV/EM parts into the last thing you see, so

 Mon Jul 04 16:42:22 +0000 2022

god i love POSIX phrasing

 Mon Jul 04 17:20:07 +0000 2022

in just enumerating du flags ive already encountered a classic coreutils lie moment: https://bugs.debian.org/1014357

 Mon Jul 04 19:53:01 +0000 2022

twofer https://bugs.debian.org/1014360

 Mon Jul 04 20:57:42 +0000 2022

threefer AND consecutive? oh my https://bugs.debian.org/1014361

 Mon Jul 04 21:50:47 +0000 2022

turns out elaborated-type-specifiers do NOT sit well with pointers-to-member, christ

 Mon Jul 04 22:52:43 +0000 2022

gdb-induced momentosis

 Tue Jul 05 02:14:46 +0000 2022

/and/ a free mandoc parsing bug! https://inbox.vuxu.org/mandoc-discuss/20220705151536.tqmlsw4z4pqraue6@tarta.nabijaczleweli.xyz/T/#u

 Tue Jul 05 15:21:28 +0000 2022

y'know what just occurred to me; this usage string for du is 450 bytes. more than 20 /entire/ programs from V1 (represented here by unix72)

 Tue Jul 05 18:03:19 +0000 2022

when minecraft posters use the third-person reverse camera and talk? thats vtoob

 Tue Jul 05 18:14:59 +0000 2022

du --time writes in date(1) format, so ive been writing the date(1) date formatter (which is similar but different from strftime(3) because fuck you), so ive found a coreutils date bug instead: https://bugs.debian.org/1014436

 Wed Jul 06 04:06:58 +0000 2022

all i want for christmas is farsi proficiency (proving santa isnt real NOT CLICKBAIT)

 Wed Jul 06 14:00:14 +0000 2022

hm

 Wed Jul 06 14:33:03 +0000 2022

king

 Wed Jul 06 17:07:03 +0000 2022

any grass touchers?

 Wed Jul 06 18:31:44 +0000 2022

how the fuck am I SIGILLing. my brother in christ clang wrote the code

 Wed Jul 06 22:04:37 +0000 2022

okay its a segfault with -O3 at least. what the fuck is clang generating that it /SIGILLs/ then???????

 Wed Jul 06 22:06:32 +0000 2022

huh

 Wed Jul 06 22:11:00 +0000 2022

bruh

 Wed Jul 06 22:24:57 +0000 2022

ok if i remove a [&]{}() IIFE wrapper it works. im gonna cry

 Wed Jul 06 22:33:15 +0000 2022

notably: I replaced /all/ the non-PODs in the IIFE with manual memory management before trying this

 Wed Jul 06 22:38:20 +0000 2022

(1) crashes, (2) doesn't; the only difference is, indeed, those three commented-out lines

 Wed Jul 06 22:42:21 +0000 2022

oh my fucking god i didnt have a return at the end. why did neither clang nor gcc warn me. im so mad

 Wed Jul 06 22:43:42 +0000 2022

ok i thought it was because i didnt have -Wall -Wextra but thEY ALSO DONT WARN. WAHT THE FUCJK

 Wed Jul 06 22:45:32 +0000 2022

am i losing my goddamn mind

 Wed Jul 06 22:47:30 +0000 2022

for FUCK's sake. i specced -isystem instead of -I. fun fact: that code /has a forbidden narrowing conversion in {}/, a hard error, that -isystem also covered up

 Wed Jul 06 22:50:57 +0000 2022

i hate strftime. in year -90 for %010F im getting 0-90-mm-dd instead of -090-mm-dd like you'd expect (well, from date(1)). for %04z im getting 000+0200 (?!??!??!?!) instead of +200 (like date(1), but thats also the only sensible output)

 Thu Jul 07 00:31:48 +0000 2022

I HATE COREUTILS I HATE COREUTILS

 Thu Jul 07 00:32:11 +0000 2022

note how this shit doesn't even degrade gracefully, right. all :-quantity variants of %2O:::z are literally, unquestionably, plain wrong

 Thu Jul 07 00:51:41 +0000 2022

https://bugs.debian.org/1014497

 Thu Jul 07 00:52:09 +0000 2022

after literal years i got signed out of android pisscorp and its apparently impossible to sign in if you don't have play services now, even if you haven't updated the app once :)

 Thu Jul 07 13:46:02 +0000 2022

lmfao glibc strftime() domain error; tm_year is the year /minus 1900/: [-2147481748; 2147485547] are i32 + 1900, and coreutils date deals with that correctly; glibc, OTOH, does /not/ and does an unchecked addition(?) – the most positive and most negative times differ by 1 day :)

 Thu Jul 07 20:11:29 +0000 2022

http://bugs.debian.org/444589, apparently, reported 30 Sep 2007 for etch; 0 replies before my own

 Thu Jul 07 20:47:54 +0000 2022

15.5kB/340L of dense C++ just to paper strftime(3) over date(1); hate it here!!

 Fri Jul 08 01:55:44 +0000 2022

im like half-convinced im just cultivating peas at this point

 Fri Jul 08 15:44:15 +0000 2022

hehe

 Fri Jul 08 16:54:13 +0000 2022

accomplishing in 6 minutes what WG21 couldn't since 2017-01-30. fuck sake.

 Fri Jul 08 23:10:25 +0000 2022

another time-based du bug, apparently coreutils considers year -90 to be later than 2021? and literally swallows it when specified explicitly, defaults to epoch????

 Sat Jul 09 00:55:48 +0000 2022

what the fuck are they *doing* to this poor fucking thing

 Sat Jul 09 00:56:31 +0000 2022

(also found an unrelated tree bug while bisecting that one, because of course: https://bugs.debian.org/1014621)

 Sat Jul 09 00:58:27 +0000 2022

--separate-dirs. of course. and it's different on my rout^Wx32 machine, which hints that it's an even worse initialisation bug. delightful

 Sat Jul 09 01:10:22 +0000 2022

filtering by du got me this old classic: https://bugs.debian.org/1007777, in which an apparently-corrupted filesystem /made du SIGABRT/. the funniest thing is that du, after i wrote the date(1) formatter and parsed all options, was a 2-hour "fill in the blanks" for me. HOW is coreutils s

 Sat Jul 09 01:12:59 +0000 2022

i mean, this is a three-fold bug because -S documentation doesn't say it affects --time at all. so.

 Sat Jul 09 01:13:56 +0000 2022

americans think about the past challenge (difficulty level: impossible) https://bugs.debian.org/1014622

 Sat Jul 09 01:55:05 +0000 2022

consecutive numbers :0

 Sat Jul 09 01:56:01 +0000 2022

posted all of this under the wrong root because im thick as pigshit https://twitter.com/nabijaczleweli/status/1545572394198306816

 Sat Jul 09 02:16:41 +0000 2022

uuh, mbrtowc(3) (by returning a (character, len) pair) and mbsrtowcs(3) (by returning the next-sequence pointer) do, and they're in C99, so?

 Sat Jul 09 20:01:17 +0000 2022

the iconv family is entirely unrelated to the rest of the ones you specced – it transliterates a byte stream to byte stream, is unaffected by the current locale (well, nl_langinfo(CODESET) if you've gotta), requires full-character alignment, and uses an unrelated mechanism

 Sat Jul 09 20:11:02 +0000 2022

most importanly, you don't get actual characters out of it (also, i recall hearing that the gnulib iconv supporting code is fucking wild with corner cases/platform idiosyncracies)

 Sat Jul 09 20:12:44 +0000 2022

oh wow, even better: glibc goes absolutely fucking apeshit (returns 0 for each mbrtowc() after the initial one that eats 2 bytes; herein wc modified to write the resulting character)

 Sat Jul 09 22:01:45 +0000 2022

guy who advertises rice as "fabulous new no-grind grain!"

 Sat Jul 09 22:46:45 +0000 2022

well that's fucking embarrassing

 Sun Jul 10 01:12:55 +0000 2022

ive had it like this since i originally ported it for df on 2021-07-18. it took me literally exactly a fucking year to notice why it's off. kernel modesetting tbh

 Sun Jul 10 01:17:39 +0000 2022

MEKHANE enthusiasts really out here like "i can fix her", huh

 Sun Jul 10 02:32:01 +0000 2022

but wasnt eth already a piece of shit chain? what could they mean by this

 Sun Jul 10 19:52:00 +0000 2022

bro i fucking hate it here point-blank derives from fucking point blanc i want to sleep a thousand deaths

 Sun Jul 10 22:51:13 +0000 2022

this is my god-damn jokair moment

 Sun Jul 10 22:54:07 +0000 2022

i also had the damn fixed-width format string backward, which is also very obvious!, but I didn't notice until i spent [REDACTED] essentially re-implementing "don't do the dot if .0". kernel modesetting 100%

 Sun Jul 10 23:54:59 +0000 2022

how. HOW. my implementation, for every pattern, literally (figuratively: manual strcspn()+strspn()) strtok()s the path, then fnmatch(3)es. how could that POSSIBLY be faster than whatever bullshit coreutils does. how could it be FOUR TIMES faster

 Mon Jul 11 01:23:29 +0000 2022

im going insane

 Mon Jul 11 01:39:17 +0000 2022

(yes, coreutils du --exclude is /four times/ slower! if you give it a trivial expression!)

 Mon Jul 11 01:47:00 +0000 2022

if you permit trivial-string optimisation then, well, i'm 471.5x faster (for the cost of three lines) :)

 Mon Jul 11 02:08:45 +0000 2022

laughing-stock. https://bugs.debian.org/1014738

 Mon Jul 11 02:20:33 +0000 2022

RT @nabijaczleweli: what if the ZFS Event Daemon had drip?

 Mon Jul 11 02:25:21 +0000 2022

okay i don't get out of bed for less than 1000x but its still fun to laugh at monopolies https://twitter.com/nabijaczleweli/status/1377977095163351044

 Mon Jul 11 02:26:04 +0000 2022

mildly annoying to have iteration times of 3 hours instead of 20 seconds, [c++ joke here], but i think ive arrived at a local winner

 Mon Jul 11 16:50:05 +0000 2022

updated to current sid and it Just Worked :0

 Tue Jul 12 14:08:24 +0000 2022

the year of linux on the babtop has arrived, after only one year, six months, and two weeks

 Tue Jul 12 14:29:11 +0000 2022

finally bisected the coreutils dd data loss bug i had a comment for in my tests for close to a year now https://bugs.debian.org/1014839

 Tue Jul 12 23:08:19 +0000 2022

well, okay, I just thought it was just an unexpected interaction or whatever, but yeah nah its just a data loss moment

 Tue Jul 12 23:09:37 +0000 2022

W + ratio

 Wed Jul 13 12:08:00 +0000 2022

left: clang trunk, bullseye gcc (10.2.1-1)
right: Issue 7
why!!

 Wed Jul 13 13:33:35 +0000 2022

"but it's CX" yeah so why isn't there a flag to make the format string checker check for POSIX strings! if the only way to work around a hard false positive is to #‌pragma GCC diagnostic (-Wformat in this case) then it's a compiler bug. blergh

 Wed Jul 13 13:37:30 +0000 2022

turns out i cant fucking read (and the man-pages page for this is noodly), and speccing "%1$ *3$.1f%2$c" works on both, no warnings; still: godawful warning especially from clang (less so from GCC, whose warning is correct, but unclear)

 Wed Jul 13 14:02:25 +0000 2022

what, and I cannot stress this enoguh, the ABSOLUTE FUCK

 Wed Jul 13 19:07:59 +0000 2022

im the dumbest motherfucker on planet earth

 Wed Jul 13 19:25:40 +0000 2022

and a consecutive twofer for FreeBSD: https://bugs.freebsd.org/265199 https://bugs.freebsd.org/265200

 Wed Jul 13 20:09:18 +0000 2022

ill be honest, i didnt expect 5kg bags of flour to be so unwieldy

 Wed Jul 13 21:03:18 +0000 2022

`sudo passwd -d $(id -un)`?

 Thu Jul 14 00:32:57 +0000 2022

and you're basing this on what? this is not the case for default sudo.conf(5)/sudoers(5) and a skim over the latter doesn't reveal an obvious way of making that be the case. making the account passwordless is /exactly/ what was requested

 Thu Jul 14 01:11:52 +0000 2022

?

 Thu Jul 14 01:20:47 +0000 2022

the former is a valid concern, sure; the latter (well, not gksudo, since that got removed in stretch because apparently it was not fit for purpose, but polkit, which I can only assume this is) Just Works, and, indeed, doesn't even prompt for confirmation

 Thu Jul 14 01:46:14 +0000 2022

very long voyages, long sight-lines, relatively slow movement v/v the medium and other vessels, hull breach being the ultimate downfall of the vessel => naval

 Thu Jul 14 14:21:31 +0000 2022

do you think you blanket rising doughs so they don't wake up before you cook them

 Thu Jul 14 15:54:06 +0000 2022

how the hell did i shake a 1/15th shot?? must be a bublik deficiency

 Thu Jul 14 18:15:25 +0000 2022

linux integer parsing moment lmao

 Fri Jul 15 11:18:04 +0000 2022

i was hunting for an overflow bug, but just found a horrible lie instead: https://bugs.debian.org/1014971

 Fri Jul 15 13:51:38 +0000 2022

mfw a distribution image has the root account passphrased

 Fri Jul 15 16:25:37 +0000 2022

i favour single-user mode or passwd -R but yeah

 Fri Jul 15 16:43:53 +0000 2022

anyone taking bugs for V7 du? (represented here in its kingest of forms – V7/x86 https://www.nordier.com/#v7x86)

 Fri Jul 15 16:54:06 +0000 2022

spot the bug (4.3BSD-Reno)

 Fri Jul 15 23:41:18 +0000 2022

fuck abraham and his prophecies, im investing all my deicoin into acyclovir

 Fri Jul 15 23:48:06 +0000 2022

fucking shameful

 Sat Jul 16 21:35:22 +0000 2022

WHY does that university site not resolve im gonna CRY

 Sun Jul 17 17:06:07 +0000 2022

fuck this. fuck it all. im done

 Sun Jul 17 17:09:56 +0000 2022

classic standard moment in XPG2 land. writes in 512-byte chunks except when it doesn't. hides errors sometimes. if you suspect you should get an error, try -r? may or may not actually reflect the space used. fuck you.

 Sun Jul 17 19:24:56 +0000 2022

at least it's actually marked PI (The behaviour cannot be guaranteed to be consistent); but this is all in addition to OF (Output format incompletely specified), because it isn't, and UN (Possibly unsupportable feature) because fucking lmao try seriously considering "-r but not b

 Sun Jul 17 19:27:47 +0000 2022

okay okay im like Reading this and i think this doesnt actually.. say anything. also note that the file in -a description is italicised, so it means "for each argument". as-specced, this is literally just a hash function

 Sun Jul 17 19:41:07 +0000 2022

i think this is all it actually says

 Sun Jul 17 19:54:42 +0000 2022

at last: preprint

 Sun Jul 17 22:10:09 +0000 2022

i hate phototypesetters I HATE PHOTOTYPESETTERS

 Sun Jul 17 22:26:05 +0000 2022

fun moment. coreutils rm -d just straight up refuses to remove an unreadable directory, doesn't prompt if -i, and lies that it tried to. what the FUCK is happening

 Mon Jul 18 19:23:36 +0000 2022

https://bugs.debian.org/1015273

 Mon Jul 18 19:37:02 +0000 2022

l + bublik + ratio

 Tue Jul 19 13:27:48 +0000 2022

i just wanted to do timezones portably and found a gcc bug instead https://gcc.godbolt.org/z/6a87eonW1

 Tue Jul 19 17:36:57 +0000 2022

hm, this says "may" so this is legal. boo http://eel.is/c++draft/temp.res#general-6

 Tue Jul 19 18:59:49 +0000 2022

i just wanted to do timezones portably but had to write the stinkiest C++ possible http://coliru.stacked-crooked.com/a/591b8f491b813c13

 Tue Jul 19 20:04:57 +0000 2022

i love freebsd

 Tue Jul 19 23:28:43 +0000 2022

https://bugs.freebsd.org/265319

 Tue Jul 19 23:46:23 +0000 2022

i love illumos

 Wed Jul 20 02:23:53 +0000 2022

the Topicbox "Join the Conversation" button says joining will let me post, but it also 250 2.0.0 Queued as 8223A064D31 so who knows

 Wed Jul 20 02:35:28 +0000 2022

appx. 100g of flour

 Wed Jul 20 11:41:40 +0000 2022

third-party analysis (1) tells me im reaching levels of slavic heretofore unseen

 Wed Jul 20 12:28:58 +0000 2022

can someone PLEASE explain to me why ext4 truncates file times at 0x37FFFFFFF

 Wed Jul 20 20:23:13 +0000 2022

for packing. documented so its fine or whatever

 Wed Jul 20 20:36:47 +0000 2022

it appears my cow cosplay has been put to rest, so this must've been some sort of additive; bublik enjoyers stay winning

 Wed Jul 20 22:44:08 +0000 2022

yeag

 Wed Jul 20 23:16:10 +0000 2022

top 10 salty spelt baker moments

 Fri Jul 22 22:54:35 +0000 2022

guy who considers tights blackleg

 Fri Jul 22 22:55:34 +0000 2022

hm. in the seventies, AT&T was the largest computer game manufacturer and distributor by far

 Fri Jul 22 23:54:51 +0000 2022

i realise this is like an insane take right but i think its also literally true by any metric?

 Fri Jul 22 23:58:38 +0000 2022

given the givens (<title>Impersonate Page), so we think this is a correct namecheap e-mail or nah

 Sat Jul 23 13:51:00 +0000 2022

its valid what the fuck

 Sat Jul 23 13:51:54 +0000 2022

both the text/plain and text/html versions are identical, yes; also: maybe don't have a title of "Impersonate Page"; also: the text/plain version of this has entirely stripped links, which is worse (the text/html has "our website" linked)

 Sat Jul 23 14:04:26 +0000 2022

okay it went through. borderline ecstatic i have no business relationship with namecheap anymore

 Sat Jul 23 14:12:01 +0000 2022

another POSIX bug (left: XBD, 5 ("File Format Notation"); right: XCU, 3, cksum)

 Sat Jul 23 19:16:12 +0000 2022

https://www.mail-archive.com/austin-group-l@opengroup.org/msg09964.html

 Sat Jul 23 19:36:51 +0000 2022

existing practice says i can output tabs. hog. https://www.mail-archive.com/austin-group-l@opengroup.org/msg09965.html

 Sat Jul 23 21:11:44 +0000 2022

:/

 Mon Jul 25 00:38:44 +0000 2022

preprint: (1); lie: https://bugs.debian.org/1015996

 Mon Jul 25 03:06:17 +0000 2022

we'd need BRISTOL STOOL CHART 1..7 to combine with PILE OF POO to accurately scale emoji prefixes imo

 Mon Jul 25 17:42:49 +0000 2022

fresh! (3 years old, added in 8.31 (2019-03-10)) coreutils env bug(?)

 Tue Jul 26 04:36:15 +0000 2022

gamer moment https://bugs.debian.org/1016049

 Tue Jul 26 04:49:33 +0000 2022

sure, and I even agree that it's useful, but the documentation describes the format as -b-s[=s[,s]...] (via a clumsy synopsis and prose, cf. the debian bug), which is not satisfied by the empty string; -b-s[=[s[,s]...]] could be =SIGS, "SIGS is a c.-std list of signal names li…"

 Tue Jul 26 11:59:39 +0000 2022

:v

 Tue Jul 26 20:02:29 +0000 2022

my stupid ass forgot the second egg once and thats why they werent fucking rising. anyway: bublik + ratio

 Wed Jul 27 21:28:30 +0000 2022

no, but only because live fish don't sediment out

 Wed Jul 27 22:56:03 +0000 2022

but then they're dead fish so they're not in the aquarium as-described. otoh live fish do shit and the shit is outside of the fish even while it's inside so even then its shaky

 Wed Jul 27 23:25:01 +0000 2022

pinky print :0

 Thu Jul 28 20:24:40 +0000 2022

i hate the gnu compiler collection. I HATE THE GNU COMPILER COLLECTION

 Fri Jul 29 14:36:18 +0000 2022

cringe-ass nae-nae baby https://gcc.godbolt.org/z/xY8zqe36T

 Fri Jul 29 14:37:54 +0000 2022

also note how the gcc error says 'static' and it very much is not literal static

 Fri Jul 29 14:48:14 +0000 2022

lovely. love that. big fan. its an anonymous namespace thing https://gcc.godbolt.org/z/5rvnPdE8j

 Fri Jul 29 15:09:26 +0000 2022

no fucking way

 Fri Jul 29 17:20:01 +0000 2022

gamer moment

 Sat Jul 30 03:08:01 +0000 2022

:0

 Sat Jul 30 16:02:01 +0000 2022

i go to bake one set of bread and now "doing what you're paid for" is "quiet quitting" i hate yanks

 Sat Jul 30 16:03:20 +0000 2022

last one babey https://twitter.com/nabijaczleweli/status/1541705431004971008

 Sun Jul 31 00:45:09 +0000 2022

im in agony

 Sun Jul 31 00:45:17 +0000 2022

how the fuck do babies do this now that we figured out blood thinners give them brain damage

 Sun Jul 31 00:47:30 +0000 2022

okay i guess they do scream 24/7

 Sun Jul 31 00:47:57 +0000 2022

im also like 2kg down since this tooth started lmao

 Sun Jul 31 12:34:55 +0000 2022