last monotone-commits post 2010-10. but! monotone-devel thread from 2024: https://lists.nongnu.org/archive/html/monotone-devel/2024-02/msg00000.html
holy hell they made PHP but for perl?? https://git.sr.ht/~nabijaczleweli/ossp-eperl
dont worry they also made an actual hell
OSSP due - Dynamic User Environment
OSSP due is a unified and dynamic GNU Bash shell and Vim editor user
environment providing reusable functionalities which proofed to be
useful in practice.
https://git.sr.ht/~nabijaczleweli/ossp-due
I had to read through all of these to come up with repository descriptions, so here's some more ones i liked:
OSSP sugar -- The Markup Language With Invisible Syntactic Sugar
OSSP sugar is a markup language and corresponding processing tool for
writing technical documentation that uses a mostly invisible markup
language (so-called "syntactic sugar" in compiler construction folk
terminology).
bro WHAT are you TALKING about (this actually appears to be an early markdown/rST-style system: https://git.sr.ht/~nabijaczleweli/ossp-sugar/tree/5adc722e/item/srml2sxml#L10)
OSSP svs - Stupid/Silly/Simple Versioning System
its incredible what mfs did to themselves before they had git, also
case "$cmd" in
v|vi|vim | e|em|ema|emac|emacs )
always execs ${EDITOR-vi} (https://git.sr.ht/~nabijaczleweli/ossp-svs/tree/trunk/item/svs.sh)
OSSP val - Value Access
OSSP val is a flexible name to value mapping library for ISO-C
variables. It allows one to access ISO-C variables through name
strings, [...]
what the fuck is an "ISO C variable" (this is actually a std::map<string, any>, but not very good https://git.sr.ht/~nabijaczleweli/ossp-val/tree/trunk/item/val.h)
OSSP cfg is a ISO-C library for parsing arbitrary C/C++-style configuration files
...parsing what? this appears to actually be a grammar compiler, unclear how this is configuration-related https://git.sr.ht/~nabijaczleweli/ossp-cfg/tree/trunk/item/cfg.pod
this also makes me the upstream for src:eperl, src:iselect, and src:osspsa. 4/54 after 20 years is pretty good though
of COURSE this shit is not UTF-8 lol
i have the usual Jörg, of course, but the remark on 81 (http://cvs.ossp.org/tktview?tn=81,4) expresses '—' as 0x97 somehow?
lol, file(1) guesses ISO-8859 but this website is somehow CP1252
aand i've done more upstream development than anyone in the past 15 years lol https://todo.sr.ht/~nabijaczleweli/ossp/205
I've instrumented the shart list importer
Error ("<20020828091907.W2689@canonware.com>") reading In-Reply-To: mail: missing '<' in msg-id
Error ("<20020730130258.A71796@canonware.com>") reading In-Reply-To: mail: missing '<' in msg-id
<20020703134036.GC1464@dt4.dev.de.cw.net> unknown charset: unknown charset: message: unhandled charset "iso-8859-1"
Error ("<20020523115704.D22880@dev14.dev.de.cw.net>") reading In-Reply-To: mail: missing '<' in msg-id
and indeed,
> In-Reply-To: <3D6BE829.CEEA7B8C@packetdesign.com>; from archie@packetdesign.com on Tue, Aug 27, 2002 at 01:59:21PM -0700
bruh.
the ISO8859 encoding thing is much worse because it doesn't affect like 5 mails and i can edit the spool manually to turn "; from" into "(from)", it affects 5000
amazing, anonymously importing "github.com/emersion/go-message/charset" automatically decodes non-UTF-8 mails
and it seems like the listssrht api program already does this(?)
i think i was hitting some internal timeouts with the big ossp-cvs spools, current status: csplit /^From / {*}; for f in *; do curl .../import -F spool=$f
if hut(1) exposed an endpoint that posted the spool i'd be using it. alas! "copy as curl" to the rescue
i may be the first and only power-user of SourceHut, the everything app
time to install alpine and sourcehut (meta and lists): 60 minutes (this is my third time (they keep breaking my old VMs))
had to do this because i managed to import /most/ of these, but not /all/
log says
2024/09/05 01:25:59 "POST http://127.0.0.1:5106/query HTTP/1.0" from 127.0.0.1 - 200 40B in 19.380792ms
2024/09/05 01:25:59 Error importing message: pq: invalid byte sequence for encoding "UTF8": 0xe4 0xf6 0xfc
2024/09/05 01:25:59 Attempt 1/1 failed (panic: pq: Could not complete operation in a failed transaction), retrying in 2m0s
and log.Printf("envelope=%q", hex.EncodeToString([]byte(envelope.String()))) says that the envelope itself has raw ISO-8859-1 äöüß, instead of =AF escaping them. so yeah that's a sane take
i should've just iconv -f iso-8859-1 | s/iso-8859-1/utf-8/. ah well. already imported :) https://lists.sr.ht/~nabijaczleweli/ossp-cvs
gamers... what the fuck is shtool versioning. no okay i get "what" it is (shtool manufactures a header file, then the build system constantly runs shtool to parse it back instead of defining the version in autohell). but why is it. https://git.sr.ht/~nabijaczleweli/ossp-uuid/commit/104b1c11e52deac0632b52337dd2f79928ce0f15
hee hee hee
[trunk 814fa05] Replace inline MD5 and SHA1 implementations with libmd
9 files changed, 42 insertions(+), 1119 deletions(-)
delete mode 100644 uuid_md5.c
delete mode 100644 uuid_md5.h
delete mode 100644 uuid_sha1.c
delete mode 100644 uuid_sha1.h
$ git diff UUID_1_6_2 --stat
54 files changed, 1520 insertions(+), 5924 deletions(-)
by my count, https://bugs.debian.org/757278, https://bugs.debian.org/864530, https://bugs.debian.org/881000, and https://bugs.debian.org/1041542, and like 3 different bugs I would consider Bad and also UUID v6/v7/max implementation and a general unfucking (i fall asleep before hitting enter yesterday)
uuid clone(); is "crash (or produce a broken UUID if it doesn’t)" because it runs the const void* constructor, which reads 16 bytes from *this (and uuid is pointer-sized)
i already had to install backports go to debug the list archive import, i dont want to learn go more than "moving hunks around" for this
i had to learn more go for my debbug (who knew someone would want to codify C-style inheritance and then call it something non-obvious like "embedded fields") https://lists.sr.ht/~xenrox/hut-dev/%3C8e0c536f50a5e6ee27bd43de3f4b854da121c0fe.1725657594.git.nabijaczleweli@nabijaczleweli.xyz%3E
had to translate the remaining manuals from pod to mdoc because pod2man generates broken pages lol; but, be hold: https://srhtcdn.githack.com/~nabijaczleweli/ossp-uuid/blob/man~000/ossp-uuid.pdf
New upstream for OSSP software is https://sr.ht/~nabijaczleweli/ossp, notice at https://lists.sr.ht/~nabijaczleweli/ossp/%3Cdas6vdk2hnazoi6b3wf5bbctpcqnqfky2nmdcwco7lkv2drmpk@tarta.nabijaczleweli.xyz%3E :)
impressive turnaround from the ftpmasters, src:osspsa
"FreeBSD ports call OSSP software »LICENSE= MIT« (&c.), but OSSP software isn't MIT (&c.)", in which a significant proportion of OSSP software in the FreeBSD ports tree is misattributed: an epic for the ages https://lists.sr.ht/~nabijaczleweli/ossp/%3Cbyjcsfzkijbtq6gsttwtffbnmtxureozobxd3xyrbgig5qsio2@tarta.nabijaczleweli.xyz%3E
well you can suck my ass and keep your archive illegal i suppose
> Hi, this is the Mlmmj program managing the <freebsd-ports@FreeBSD.org>
> mailing list.
>
> The message from <nabijaczleweli@nabijaczleweli.xyz> with subject "FreeBSD
> ports call OSSP software "LICENSE= MIT" (&c.), but OSSP software isn't
> MIT (&c.)" was unable to be delivered to the list because you are not a
> list subscriber.
in a continued trend of programmer illiteracy, gentoo says ossp-uuid is ISC (it's not)
fedora also calls ossp-uuid MIT (https://src.fedoraproject.org/rpms/uuid/blob/aa05b8cceb71977f6fc664f646456cf18f8b20fc/f/uuid.spec#_11). it's still not MIT
Removed:
* The mandate of heaven for the C++ binding
(cb6aac7aafdac8844a064fa79b1a660661c498d4)
$ git diff --stat UUID_1_6_2..UUID_1_6_3
91 files changed, 3279 insertions(+), 10289 deletions(-)
should've posted this ITS before i started archiving OSSP, it'd be half-way done by now :/ https://bugs.debian.org/1081555
i think ossp-iselect is undistributable without at least repacking since it has "all rights reserved" files
what was in the air in 1997 bro
/* define the boolean values in a general and
* portable way.
*/
#undef TRUE
#define TRUE (0 || !(0))
#undef FALSE
#define FALSE (!(TRUE))
/* typedef enum { false = FALSE, true = TRUE } bool; */
/* some other names for true and false */
#define YES TRUE
#define NO FALSE
#define GOOD TRUE
#define WRONG FALSE
(0 || !(0)) is fucking psychopathic work
I think iselect is undistributable (inlines kpathsea code, kpathsea is LGPL2.1+, iselect is GPL2)
not used, so, uh, saved? but.
dude what was happening in 1997. what does #define ZERO 0 mean. while(ZERO) is not using the 0 as fucking zero, it's using it as a false. that's the only usage btw
and the ONLY function that uses this has CUS: return rc; so this whole CU(123) machinery is just return 123;
fuck this im replacing the whole parser
https://todo.sr.ht/~nabijaczleweli/ossp/209
https://todo.sr.ht/~nabijaczleweli/ossp/210
https://todo.sr.ht/~nabijaczleweli/ossp/211
https://todo.sr.ht/~nabijaczleweli/ossp/212
https://todo.sr.ht/~nabijaczleweli/ossp/213
https://todo.sr.ht/~nabijaczleweli/ossp/214
(i was already replacing the whole parser. this is just what I found while figuring out what my new parser should do)
god i keep reinventing urlview. motherfuckers just love urlview
^D isn't DELETE you psychopath what the hell??
if (c < ' ' || c > '~') {
/*
* a control code
*/
if (c == '\n') { /* RETURN */
}
else if (c == 0x1b) { /* ESCAPE */
}
else if (c == 0x04) { /* DELETE */
}
else if (c == 0x7f) { /* BACKSPACE */
i fixed the parser. i removed every length limit. i support embedded NULs. it allocates no additional memory beyond the file data (used to be 2x at least). it's ready for wide characters. it's more reentrant (i think?). it leaks less memory.
+261 -297
@nabijaczleweli yeah, the nerve of people insisting we have to subscribe, register an account, … just to report a bug… Debian does it right.
@nabijaczleweli It looks like MIT to me, despite not seeing it explicitly mentioned as such? At least, with the uuid-1.6.2 tree from ossp.org. The license page on the site also seems to agree. Not sure where the claim of ISC could come from, they use GPL/LGPL 2, MIT or BSD only...
@nabijaczleweli oh nevermind I see :)))
@astraleureka oh wow, i didnt even notice the licences page... yeah, its crazy that https://lfs.nabijaczleweli.xyz/0022-OSSP.org-mirror/ftp.ossp.org/ossp-web/doc/license.html calls rse's funny OSSP licence MIT, and rse's funny four-clause BSD-style licence "The BSD Licence" (freebsd ports calls it BSD-4-Clause (it isn't)). i just thought he pasted some chunks together for a bespoke momento, but this reads like he thought those were the MIT/BSD licences? where could he get this idea(/text?) from
@astraleureka honestly the best match for rse's funny OSSP licence is 0BSD i think (but 0BSD post-dates rse using his by 5+ years)
@nabijaczleweli *mongols rush in*
@the_art_of_giving_up omg thats me im the mongols in this case
@nabijaczleweli thank you for doing this
@nabijaczleweli oh no, that's not allowed? why does anyone put up with these licenses....
@leftpaddotpy apparently it is (both embedding and relicensing) so iselect seems distributable from this angle at least
@nabijaczleweli LGPL2 can go into GPL2