<testuser[m]>
I think dilyn meant toybox ls hant got colors
<testuser[m]>
Hasn't
<claudia>
this is meant for toybox
<claudia>
_my_ toybox ls got color.
<testuser[m]>
Oh nice
<schillingklaus>
i do not want any colours in the shell
<claudia>
when building with toybox, I got a bunch of 'mv/find no such file' after extracting the source but the actual build goes just fine. http://ix.io/3tGe
<schillingklaus>
claudia charmed toybox into becoming colourful
<claudia>
surething
<soliwilos>
cem: I built your otools 1.4.1 package and unless I did something wrong, doas is installed without setuid bit. For my local fork I just added a chmod in post-install, but you may want to fix it in your package?
<schillingklaus>
otools == tools from openbsd userspace?
<soliwilos>
Yes.
<cem>
soliwilos: that's weird
<cem>
Makefile does that
<soliwilos>
Only some of them, though. Not sure if the intent is to add more to it over time.
<cem>
And I have no problem with my package
<soliwilos>
Hmmm.
<soliwilos>
The only thing I changed was the libressl dep.
<cem>
Try adding a chmod u+s on the package build script
<cem>
And remove the post-install
<cem>
If it isn't setuid that means kiss fucks up permissions
<soliwilos>
Doesn't look like there's any chmod commands in the kiss repo opendoas package.
<soliwilos>
Well, not on the doas binary.
<cem>
Yeah, because both Makefiles do the chmod themselves
<cem>
I'm assuming that kiss accidentally or intentionally strips setuid
<soliwilos>
It's weird then that otools wouldn't install with. Hmmm, could it have to do with kiss's alternatives feature? Seeing as I already had opendoas installed, doas from otools was put in /var/db/kiss/choices..
<dilyn>
odd that kisslinux/repo doesn't use it tho
<testuser[m]>
dilyn have you tried building chromium in tmpfs
<testuser[m]>
it takes 1 min for ninja to even start on hdd lol
<dilyn>
it builds in tmpfs if I decrease the job numbers
<dilyn>
otherwise i OOM
<dilyn>
tmpfs also has to be > 9GB lmfaoooo
<dilyn>
otherwise it won't even extract
<testuser[m]>
cuz kiss needs a tmp tar
<testuser[m]>
i think the ram requirements could be reduced a lot by removing most stuff from third_party/ and chromeos/
<dilyn>
oh actually i think that got removed recently!
<sad_plan>
ah, didnt see that patch. I did actually look at the commitlogs for firefox in wyverkiss, just to make sure I didnt miss anything, but it seems I did anyway. Ill add the patch and try again. thanks
<dilyn>
so now you just need 5GB lol
<testuser[m]>
no it was reverted
<dilyn>
f
<dilyn>
the moral to the story here is to just use pax
<cem>
is there a saner gdb alternative?
<testuser[m]>
lldb ? not saner though
<dilyn>
sed -i '/main()/a break;/' foo.c :V
<cem>
lmao
<testuser[m]>
btw wdym saner ? complexity wise or usage ?
<dilyn>
gdb is very opaque
<cem>
both, i guess
<sad_plan>
the patch fails. it seems the dir does no longer exist for some reason
<sad_plan>
wait
<sad_plan>
ah, the file its trying to patch, is changed, so its missing with a couple of lines
<sad_plan>
noow its building :D
<sad_plan>
they actually upstreamed parts of the patch
<konimex>
Hmm I think I already removed the if_indextoname patch in wyverkiss since 90.0
<sad_plan>
yep, you did
<sad_plan>
however, the first line is not changed upstream, just the second part, which probably is the part that fails on me. ill see in about 40 minutes or so,, when its gonna build libxul again
gtms has quit [Remote host closed the connection]
micro_O has joined #kisslinux
<sad_plan>
can you not build a statically linked llvm with lto? I was reading about lto on llvm, and it refered to liblto as shared libs. said nothing about static. however, I was looking at dilyn commit log for his llvm build, and noticed he disabled lto when he built it statically. same with pgo(?)
<testuser[m]>
Why would you want a static llvm
<testuser[m]>
200mb static library linked into clang, rust, lld, everything
<micro_O>
o/
<testuser[m]>
Hi
<sad_plan>
why not?:p
<sad_plan>
o/
<micro_O>
I love the diff from 5.5.18...5.5.17
<testuser[m]>
Cuz its a waste of 1gb
<testuser[m]>
and time
<sad_plan>
in any case, I dunno. Im just trying to build stuff statically, aaswell as enabling pgo/lto on stuff that can.
<testuser[m]>
Atleast the ver bump.should be there
<dilyn>
you can certainly build a statically linked llvm but I'm not sure how things like libLTO function so I have no idea how it impacts anything
<dilyn>
as for disabling PGO/LTO in the llvm build, you'll often see options change in my build files when I run into symbol errors. iirc, I was having some issues relating to relocation
<sad_plan>
^I was just gonna ask you. I did however noticed that you disabled lto/pgo, when you enabled static
<dilyn>
considering there are 17 commits between 5.5.17 and 5.5.18? yeah github is broken i think
<dilyn>
you should only do one fucked up thing at a time (:
<testuser[m]>
Why did github add reactions on releases
<testuser[m]>
Might as well make a comment section there too
<sad_plan>
why? borking my system, and fixing it later is fun :p it great for learning. ..or something
<sad_plan>
¯\_(ツ)_/¯
<dilyn>
breaking your entire toolchain is unpleasant xD
<acheam>
testuser[m]: yes I want to comment "first!" on a KISS release!
<dilyn>
borking basic tools also sucks :X i have come to fear fiddling with toybox. i haven't updated it in a long ass time
<sad_plan>
I knoow. iirc, last week I broke it so hard, I had to reinstall. nothing worked. I just couldnt figure out how ti fix it, like I use to.. which was honestly a bit frustrating :p
<dilyn>
these reactions are weird because they're like, all people who haven't contributed to community lmao
<testuser[m]>
Just keep another chroot open in terminal with real root bind mounted :p
<dilyn>
the absolute madlad
<sad_plan>
basic tools is fine, rather easy to fix, its when gcc/binutils just outright refuses to work, your in trouble :p
<sad_plan>
lol
<dilyn>
honestly sad_plan what you should do is set KISS_ROOT=$HOME/fuckit and do your work in there
<sad_plan>
lol absolutely
<dilyn>
you'll inherit some taint from your host system but generally this is okay
<sad_plan>
its all good :D
michalsieron has quit [Remote host closed the connection]
micro_O has quit [Ping timeout: 255 seconds]
micro_O has joined #kisslinux
michalsieron has joined #kisslinux
<micro_O>
i do like the idea of having / be read-only, and then do some bind mount rw shenanegains until ready to commit the changes wholesale
<micro_O>
like maybe theres some crazy way to have pre-merge or pre-install hook that just mounts over the new files...
<dilyn>
KIMMUTABLE
<micro_O>
kimm() { export KISS_ROOT=$(mktemp -d kimm-overlay.XXXXXX); kiss install; folder2unionfs $KISS_ROOT; mount -o unionfs / $KISS_ROOT }
<micro_O>
(having never used unionfs or the like)
<testuser[m]>
mktemp is not portable
<testuser[m]>
Off to gulag
<micro_O>
im not rewriting mktemp for muh portability for something as weird as what im proposing
<dilyn>
mkdir -p /tmp/kimm-overlay.$((RANDOM)) || printf 'use a real shell you monster\n'
<sad_plan>
dilyn: I though for a sec you were joking, but I totally forgot about the KISS_ROOT function. thats genious. Im sure its easier to work with, than just always using a tarball to mess with instead
<dilyn>
a chroot is an excellent choice when you want to avoid taint from the host, but if you're building static objects this isn't the biggest of concerns, especially if the only difference between host/target is static v shared
micro_O has quit [Ping timeout: 252 seconds]
<sad_plan>
indeed
<sad_plan>
oof. firefox failed. again..
<testuser[m]>
install chromium
<testuser[m]>
or links
<testuser[m]>
Same log ?
<sad_plan>
I have lynx actually, but yes, I am indeed considering switching browser, for real. even though I really like mye ff setup..
<sad_plan>
you mean rebuilding it from the dir where it failed?
<testuser[m]>
umm did you rip out the other half cuz it was failing ?
<testuser[m]>
Or did it say detected reverted patch
<sad_plan>
yes, I peeked at the file, and it appeared to be upstreamed
<sad_plan>
so I removed it
<sad_plan>
there was nothing to patch really. iirc
micro_O has joined #kisslinux
<sad_plan>
I can send you the file aswell, so you can verify yourself aswell
<micro_O>
dilyn, wouldn't it need to be something closer to: while test -d attempted_temp=${TMPDIR:-/tmp}/kimm-overlay.$((RANDOM)); do; done; mkdir -p $attempted_temp
<micro_O>
actually more complex, assignment doesnt echo the assigned variable
<testuser[m]>
It wasnt upstreamed
<testuser[m]>
There was nothing to upstream
<testuser[m]>
It was just a hack
<dilyn>
my line was a joke rather than a serious suggestion xD
<testuser[m]>
It's just moved behind an ifdef that is always true
<testuser[m]>
In our case
<testuser[m]>
So just make the return 1280 change here
<sad_plan>
oh well
<testuser[m]>
Why did you think it was upstreamed ?
<sad_plan>
I dunno, I must have thought it was gone for some reason. I looked at the file, and dont think I saw the content I was looking for, and thought, oh well, its probably fixed. and went merry on my way to a failed compiling
<sad_plan>
in any case, I readded that part, and it patched succsessfully. the first part however, was off by 3 lines or so, so it failed, so I had to change the line numbers
claudia has joined #kisslinux
<sad_plan>
is surf still considered really slow these days aswell? I remember viewing a video of distrotube, where he kept complaining about how immensly slow surf was, even though it didnt really appear that slow compared to the other browsers he tested in the video
<claudia>
dilyn: toybox sort is causing the error mv/find messages when extracting stuff.
<dilyn>
i would be amazed if that were the case...
<dilyn>
waht give syou this impression?
<claudia>
I swapped out every utiliti until the error was gone
<claudia>
swapping back brings back the issue
<dilyn>
what version of kiss are you using?
<claudia>
5.5.18
michalsieron has quit [Ping timeout: 256 seconds]
<msk[m]>
sad_plan: I don't recall who, but someone here said that surf ran really well for them
<micro_O>
dilyn if $_ is posix, then: faketmp() { while ! mkdir ${TMPDIR:-/tmp}/kimm-overlay.$((RANDOM)); do; done; echo $_ }
<acheam>
sad_plan: its as slow as webkit2gtk is
<claudia>
^
<acheam>
i dont think it caches anything though
<acheam>
and definitely doesn't do any link preloading or anything
<micro_O>
ok, $_ isnt in the yash documentation, so final version of mktemp replacement
<micro_O>
fktemp() { while ! mkdir "$@" 2>/dev/null; do set ${TMP:-/tmp}/kimm-overlay.$((RANDOM)); done; echo "$@" }
<ang>
RANDOM isn't POSIX
<micro_O>
ang what reference are you checking?
<micro_O>
i actually dont know the best way to check
<micro_O>
I just go by what 'yash' says is supported
<acheam>
i just know it doesnt work in ash
<ang>
man 1p sh usually
<ang>
but in this case I just know
<micro_O>
hmm, man 1p sh fails for me, but i am on a bsd-ish, not a linux-ish
<acheam>
an equivalent is "od -A n -t d -N 3 /dev/urandom"
<ang>
you probably have to install manpages-posix or equivalent
<sad_plan>
msk[m]: I belive that might be cem, iirc.
<dilyn>
claudia: i'm not sure what to make of all this. my issue immediately vanished when I updated to 5.5.18
<sad_plan>
acheam: that might be the case, and my firefox isnt inherently fast either, because the darkreader addon, is making my browser slow as a turd. im not caching anything either iirc
<ang>
finally came around to giving acme a proper try
<acheam>
then you probably wont find surf slow
<ang>
it's pretty cool for building interfaces for random shit
<ang>
not so great for actual text editing, relying so heavy on the mouse is rather annoying, especially with a track{pad,point}
<sad_plan>
ang: did you like it? im in the middle of trying to package plan9port in my repo, but its giving me some stupid issues with a directory does not exist, when it clearly does :p
<sad_plan>
acheam: probably not. gonna try it out later today, or tomorow or something
<sad_plan>
really like that its keyboard driven. installed tridactyl for firefox, so I can have vim keybindings in ff, which is amazing
<ang>
it's a neat concept but as I said, not a big fan of using the mouse
<acheam>
just remember that it isnt a minimal browser
<acheam>
its just a bloated browser without a ui
<acheam>
no real purpose for it IMO
<ang>
sad_plan, I haven't packaged it
<sad_plan>
ang: not having to use the mouse, is imo preferable tbh. only thing I use my mouse for these days is marking text really :p
<ang>
I just extracted and build it from there
<sad_plan>
acheam: is surf not a minimal browser? or are you refering to the big bloated mess of firefox? :p
<sad_plan>
aah ok
<acheam>
yep talking about surf
<claudia>
dilyn: hm that interesting and odd. Maybe its going to be a mixture of the utilities we are using.
<dilyn>
probably
<sad_plan>
well if surf isnt a 'minimal' browser, then what is? just terminal based ones?
<ang>
sad_plan: I haven't edited much text with it yet, mainly just used it as an interface for reddio
<dilyn>
I don't know why I expect good things to happen when I do this shit XD
<ang>
and that mostly involves middle clicking certain text to load new posts
<sad_plan>
msk: because the web is bloated :p
<sad_plan>
¯\_(ツ)_/¯
<ang>
sad_plan, I would definitely learn sam first. Knowing its command languange and structural regular expressions already means one less thing to learn in acme
<sad_plan>
ive been wanting to try out sam, or just the other plan9 stuff in general. their really interesting. but ill keep that in mind indeed :D
<sad_plan>
theyre*
dbz has quit [Read error: Connection reset by peer]
humky has quit [Quit: Leaving]
dbz has joined #kisslinux
<micro_O>
msk[m] depending on what you want in a browser, i think they can exist
<micro_O>
it would be a decent amount of work, and i would maybe not call it a 'browser' anymore, but a user-agent
<micro_O>
rss readers, as an example
claudia has quit [Read error: Connection reset by peer]
<msk[m]>
rss readers don't need to deal with js or css
<msk[m]>
unless the feed has something like
<msk[m]>
I guess
<msk[m]>
p style="color: red;"
<testuser[m]>
You can have a minimal browser, you just wont be able to browse the trash sites with it
<msk[m]>
I need some trash sites for school
<msk[m]>
I thought others would be in similar situations, where they would prefer not to use trash sites but sometimes have to
<sad_plan>
honestly, I find myself browsing less and less complicated sites. however, there are occations where I need to though.. appart from that, Im sure I could manage with a terminal based one
<sad_plan>
the most complicated site I browse on a regular basis, is probably github
claudia has joined #kisslinux
<acheam>
you don't need to browse that on a daily basis :)
<claudia>
dilyn: do you use pax with your kiss?
<dilyn>
mmmhmm
<sad_plan>
in reality, no. I just look at projects I find, and stuff, I suppose. alot of it I could go though my termial anyway, but yeah. sometimes its just easier to use the browser. like when looking through commits. I havent found a way I can comfortably view commits/diffs in terminal yet :p
<claudia>
Maybe this can causes the different behaviour on our sides. Mind to share the code?
<dilyn>
I thought testuser said the second patch is just uneccessary
<sad_plan>
here is the wyverkiss patch, which contains patching of both files.
<dilyn>
either way, the second file is probably just http://ix.io/3tJ7 if you super dont' care about any of the work they did
<sad_plan>
did he? I was intending to merge them afterwards
<dilyn>
i might've deleted a few too many lines in that patch i didn't read super closely lol
<dilyn>
but none of this explains why, if you're using all the latest repos, you would even need this...
<sad_plan>
I knoow. the only difference here is im still on xorg, and libressl, but that shouldnt really make any difference here, as ff doesnt have a dep on any of them really, or if any, its xorg, and I already got that obviously..
<sad_plan>
it isnt webrtc related by any chance?
<sad_plan>
I just know its disabled in the latest builds
<dilyn>
that may very well be worth a bug report...
<dilyn>
with toybox, that is
<dilyn>
E5ten: ?
Guest47 has joined #kisslinux
micro_O has quit [Ping timeout: 252 seconds]
micro_O has joined #kisslinux
<Guest47>
dilyn hey can you help with shell script
<Guest47>
Trying to read files from a directory into a variable via for loop, ONLY including files (no directory) but it's not working. It's printing directories too. https://pastebin.com/YBW2qMMP Sory for formatting
<Guest47>
alright never mind igot it
<Guest47>
had to use -d flag but it was fucking up for some reason before, don't know why it works now
<claudia>
dilyn: Got a good title for that issue?
<claudia>
sort: not behaving as it should
<acheam>
Guest47: you really don't need those braces around variables
<acheam>
also, you can rewrite the if statement using flow control operators
<micro_O>
I'm going to clean it up, publish it, and make a PR for kiss community.
<acheam>
then you do a you
<Guest47>
don't tell me what to do
<Guest47>
>:)
<Guest47>
if i dont want to do me i wont
<acheam>
micro_O: next step: make it fix things like permissions issues an open a pr
<micro_O>
acheam i do intend to add a --fix param to the linter for very clear things
<micro_O>
like permissions errors
<micro_O>
might be worth adding as a pre-push or pre-commit hook, but I don't want to dictate other folks workflows
claudia has quit [Quit: zzz]
<acheam>
you cant do that for people anyways
<Guest47>
acheam would you use 0 to mean true and 1 to be mean false in shell scripting boolean logic. that's what exit code seems to do, with 0 meaning "ok" and 1 meaning "error"
<acheam>
no its a little backwards like that
<acheam>
[ false ] && echo "hi"
<acheam>
would echo hi
<acheam>
ehr
<acheam>
wait
<acheam>
one sec let me think about thi
<acheam>
s
<Guest47>
it does echo hi i just tried it
<Guest47>
because false is equivalent to 1...
<Guest47>
but so doese [ true ] && echo "hi"
<acheam>
yeah thats whats tripping me up lol
<acheam>
ang: you in your endless shell wisdom, may you please help us out
<Guest47>
oh [ false ] just tests to see if the string is empty
<acheam>
ah
<Guest47>
So i guess it's just using 0 and 1 in opposite ways for exit codes
<Guest47>
I honestly think 0 should mean error and 1 should mean no errors
<acheam>
micro_O: nice
<acheam>
Guest47: the problem with that is then you only get 2
<acheam>
options
<acheam>
but with the other way around you have 253 choices
<acheam>
for more descriptive exit codes
<Guest47>
oh yeah you're right
<Guest47>
cause exit 2 has something to do with shell built-in iirc
<ang>
uhm, let me read the backlog
<acheam>
basically
<acheam>
why does "[ false ] && printf "hi\n"" work
<ang>
well you figured it out already
<acheam>
yeah
<ang>
false is just testing the string false
<ang>
different if you did `false && echo true`
<ang>
normally you use [ -n "$var" ] to test if a variable is empty
<Guest47>
I also found out that by having exit codes return 1 and up for more descriptive error codes. I was just thinking how this would be confusing if you're writing boolean logic
<ang>
or not empty rather
<ang>
but you can also do [ "$var" ]
<acheam>
so -n is bloat
<ang>
I guess, lol
<Guest47>
Like if you have a function is_true () { ... ; } this is very confusing because if you're following the convention it should return 0 if it is true (indicating successful execution)
<Guest47>
Normally in something like C is_true() would return 1 if true
<Guest47>
That's where I'm confused
<Guest47>
Return value in shell scripting should be the same as exit codes right? Like return 0 if there is no issue with the function execution?
<ang>
yes
<Guest47>
Okay so if you wanted to use boolean logic you'd hjave to do ! is_true ...
<Guest47>
with the exclamation mark
<acheam>
or use ||
<Guest47>
ah so is_true || ...
<ang>
I don't understand why you'd ever want a functiom is_true() and return non-zero for true
<acheam>
have you seen the gnu implementation of true?
<acheam>
there are instances where it can return nonzero
<Guest47>
I was just using is_true as an example
<Guest47>
I guess it's just a semantics issue
<ang>
haven't looled at the implementation bit I've hear it can do that
<ang>
looked*
dilyn has joined #kisslinux
<ang>
typing on phone, too many mistakes to correct them all
<ang>
no auto correction in termux ftw
<dilyn>
big fan of this kiss-lint work
<ang>
it should also look for tabs and trailing spaces
<dilyn>
so far it looks quite nice tho! ++++1 :D
<ang>
aye, good to automate something like this
<micro_O>
thank you dilyn
<micro_O>
Its a long-term stepping stone to some more general test, build, and release automation.
<micro_O>
*most* of kiss is simple enough that I do not need to fight the package manager or package format, which is really cool
<acheam>
I think we know what the rest of "most" is
<dilyn>
absolutely. This thing could be very, VERY powerful