<claudia>
I mean your shell and the unzip stuff still works?
<sad_plan>
yeah, I can still navigate the system as normal, but when building i.e busybox, itll error out telling you /usr/bin/ld is missing, even though its there. ive also tried to use kiss a with a binutils-bin package, but for some reason, no dice
<sad_plan>
ive messed up my system in all kinds of ways. cp is missing, mv is mising, grep is missing, and so on :p the tarball is always handy tbh :p
<sad_plan>
it all works now though. just /bin/ld this time
claudia_ has joined #kisslinux
claudia has quit [Ping timeout: 246 seconds]
<claudia_>
'kiss-manifest binutils', is ld there?
<claudia_>
when u said messing with cflags, what do they return now?
<claudia_>
'which ld'
<sad_plan>
kiss manifest does return /usr/bin/ld.
<claudia_>
so 'which ld' is also success
<sad_plan>
which ld returns /usr/bin/ld, yes
<sad_plan>
my cflags is nothing too special. its something in the lines of what mid has. im just seeing if I can make my system use less ram at idle or something.
micro_O has joined #kisslinux
<claudia_>
Dunno, when have messed with these I would try settings CFLAGS LDFLAGS to nothing and test.
<sad_plan>
wont that just revert to a default?
micro_O has quit [Ping timeout: 265 seconds]
<sad_plan>
in any case, setting blanks did nothing for me, same error. ld is still missing it seems. other packages will just return gcc cant produce executables
<sad_plan>
when it checks if gcc works, itll return 'no'
<claudia_>
I am not sure what default should be. I think it does setting to nothing.
<GalaxyNova>
same reason to use kiss package manager
<riteo>
ok but git uses binary formats
<acheam>
thats not actually git though
<acheam>
ddevault shit actually is git
micro_O has joined #kisslinux
kayw has quit [Quit: WeeChat 3.2]
kayw has joined #kisslinux
<riteo>
if you wanted to have a true usable shell git you might as well make a new vcs from scratch made to be compatible with unix text processing tools in the first place
<acheam>
as mid would say
<acheam>
the only good version control system is versioned tarballs of the source released over FTP at undetermined intervals
<riteo>
lmao
<riteo>
fool proof
jleightcap2 has quit [Ping timeout: 265 seconds]
<acheam>
wow theres actually a lot more perl in git than i thought there was
<riteo>
I have mixed feelings over switching to kiss in general
<illiliti>
noocsharp: that concept reminds me of wmutils
<acheam>
DYLAN: are you set on using libcurl for k?
<acheam>
idk how youre going to respond to that
<acheam>
but you have loyally responded to my other unidirectional questions and complaints, so I have faith that you will
<noocsharp>
how? lol
<GalaxyNova>
what?
<necromansy>
just open an issue :>
<necromansy>
lmao
<kqz>
he says you can shoot him an email as well
<acheam>
but that requires github
<acheam>
and I already have irc open lol
<riteo>
on one side, it'll be way more stable and fast (no more risking a system break over base tools and it will be possible to statically link the executable) but on the other it will be less portable (I think?) and possibily even bug-prone
<GalaxyNova>
please let's not depend on libcurl
<acheam>
^
<riteo>
I'm curious, what other alternatives are there?
<GalaxyNova>
riteo: Also the binary will be way larger
<noocsharp>
i mean the shell package manager already depends on curl
<GalaxyNova>
it does not anymore
<illiliti>
libfetch
<GalaxyNova>
there's KISS_DL
<kqz>
portability should be fine, it'll be posix and most likely statically built
<acheam>
no that was removed
<GalaxyNova>
you can use wget in kiss
<acheam>
but its still trivial to replace
<acheam>
mid and I both use Axel instead of curl
<acheam>
maybe a configure flag to let you chose between libcurl or system(utility of your choice)
<GalaxyNova>
I was about to mention system() too
<noocsharp>
i think the bigger question is how will git be handled
<noocsharp>
in my attempted c package manager, i stole an http implementation from hurl from codemadness.org
<acheam>
got uses git under the hood
<acheam>
and doesnt have an easy way of amending commits
<noocsharp>
seemed to work well enough
<acheam>
ofc hiltjo has a solution
<GalaxyNova>
what about readelf?
<testuser[m]>
It could be stolen from binutils
<noocsharp>
libelf?
<testuser[m]>
the implementation
<GalaxyNova>
ok
<illiliti>
it could be written from scratch because elf format is easy to parse
<GalaxyNova>
I just can't see any major benefits that would justify rewriting the package manager in C
<noocsharp>
speed
<illiliti>
statically linked binary
<GalaxyNova>
the main bottleneck kiss has now is the readelf, that won't change with C
<GalaxyNova>
and what's the point of statically linking the binary if the resulting binary will still be larger than the current one
<testuser[m]>
sh is static anyway
<testuser[m]>
On mosr KISSes
micro_O has quit [Ping timeout: 268 seconds]
<GalaxyNova>
but you need sh for the system to work
<GalaxyNova>
so it's a passive dependency
<riteo>
also
<noocsharp>
you need a c compiler too...
<riteo>
you need a lot of baseutils
<necromansy>
we *need* a c compiler atm anyway tho
<testuser[m]>
Btw why the talk of the C package manager ? Did dylan put out something yesterday ?
<illiliti>
kiss depends on coreutils. if coreutils breaks, the entire system becomes unrecoverable
<riteo>
not really, it was a very old project
<riteo>
illiliti: not necessarily
<GalaxyNova>
testuser[m]: the kiss package manager is now over 1000 lines of code
<noocsharp>
there are a lot of things that can be done more efficiently in c than in shell
<noocsharp>
beside readelf
<riteo>
illiliti: I mean, if the package manager was statically built from c
<testuser[m]>
The biggest benefit would be the kiss won't break with alternative coreutils implementations
<GalaxyNova>
kiss shoudn't break with alternative coreutils implementations if the implementations are also POSIX
<GalaxyNova>
compliant
<testuser[m]>
Most of those issues have been fixed but a few still pop up from time to time
<dilyn>
dylan can't use a statically linked K *and* libgit2 because then K is GPL :P
<riteo>
BRUH
<riteo>
libkgit suddenly appears
<dilyn>
I am unconcerned with kiss being slightly over 1k lines because a fair few dozen lines are fallback/redundant/can-be-removed, or syntactic choices
<dilyn>
ideally what dylan would do is develop some way to use an arbitrary source download method selected at compile time
<dilyn>
then he wouldn't have to worry *too* much
<GalaxyNova>
still, it open a gate for more feature creep
<dilyn>
ideally that choice is curl
<dilyn>
'feature creep' it's a package manager dog
<dilyn>
there are like, eight features
<necromansy>
also more loc > obfuscation and PM de-effieciency
<riteo>
dilyn: I thought too that a package manager would be comples
<riteo>
s/comples/complex/
<riteo>
then I started writing minekiss
<dilyn>
Most things now are relegated to their own functions and a lot of changes have been documented pretty well, so I don't think kiss has become any harder to read
<dilyn>
harder to keep track of certainly, but that's inherent in a shell script past 100 lines
<acheam>
minekiss has shaped the man riteo has become
<dilyn>
lmao
<riteo>
lmao indeed
<riteo>
I actually learned quite a lot and I still have so much room to improve it
<riteo>
like, the biggest lesson I learned is: put as little logic as possible into your script
<riteo>
it sped up A LOT the script
<dilyn>
it should always do the minimum
<riteo>
like, the naive approach took 2:30 minutes only to list the asset names
<riteo>
now it's way faster, actually usable
<riteo>
I might even port it to C one day just to learn it, although minecraft's version format is pure partially undocumented hell
<dilyn>
as far as 'kiss requiring foo', you have to remember that this is a source based distribution with a statically linked set of core utilities. busybox, curl, and git are all statically built. if you fuck up your toolchain, that isn't KISS' fault
<dilyn>
you can embed kiss in busybox itself if you want
<dilyn>
nothing pertaining to the actual functioning of the system should ever be able to break because you can almost ALWAYS recover
micro_O has joined #kisslinux
<illiliti>
initramfs saved my ass many times ...xd
<dilyn>
right? dropping to an emergency shell is very powerful
<GalaxyNova>
just use a chroot
<dilyn>
especially when initramfs includes the entire busybox binary lmfao
micro_O has quit [Ping timeout: 265 seconds]
micro_O has joined #kisslinux
zola has quit [Ping timeout: 252 seconds]
micro_O has quit [Ping timeout: 265 seconds]
micro_O has joined #kisslinux
micro_O has quit [Ping timeout: 255 seconds]
GalaxyNova has quit [Read error: Connection reset by peer]
akira01 has quit [Remote host closed the connection]
<konimex>
> dylan explicitly wanted to avoid it, it requires cmake to build
<konimex>
as if that's a bad thing
<konimex>
(at least from a GNU-free perspective)
protonesso has joined #kisslinux
<testuser[m]>
Makefile
Uks2 has quit [Ping timeout: 258 seconds]
Uks2 has joined #kisslinux
GalaxyNova has joined #kisslinux
zola has joined #kisslinux
zola is now known as Guest8513
Guest8513 has quit [Ping timeout: 265 seconds]
zola has joined #kisslinux
zola is now known as Guest2417
schillingklaus has joined #kisslinux
Guest2417 has quit [Ping timeout: 268 seconds]
GalaxyNova has quit [Read error: Connection reset by peer]
mrlix has joined #kisslinux
<riteo>
gtg, bye!
riteo has quit [Quit: epic minekiss shaped me into who I am moment]
rio6_ has quit [Quit: WeeChat 3.0.1]
illiliti has quit [Quit: leaving]
Uks2 has quit [Quit: Byee]
Uks2 has joined #kisslinux
michaelskyba has quit [Ping timeout: 265 seconds]
Uks2 has quit [Ping timeout: 258 seconds]
Uks2 has joined #kisslinux
f1nniboy has joined #kisslinux
soliwilos has joined #kisslinux
micr0 has quit [Ping timeout: 268 seconds]
Uks2 has quit [Quit: Byee]
Uks2 has joined #kisslinux
lolololol has joined #kisslinux
lolololol is now known as sad_plan
<sad_plan>
why on earth cant I get permission to open /dev/null? Im doing a wyverkiss install, and when I run kiss u, I keep getting the error telling me fata: open /dev/null or dup failed: permission denied
<sad_plan>
im root btw.. but same issue if I run as a regular user
<sad_plan>
tried both an older tarball aswell.
<testuser[m]>
ls -l /dev
<testuser[m]>
does :>/dev/null work
<soliwilos>
I've neither tried wyverkiss or seen that problem before, does /dev/null exist?
<sad_plan>
yes :> /dev/null works. ill give you the other output in a sec
<dilyn>
dylan made a new release that should include all of the fixes related to .cache problems
<dilyn>
5.5.6 is the latest that a tag exists for, though 5.5.7 has been made (just without a tag)
<msk[m]>
I updated the package manager just today
<noocsharp>
wait, the build file is using $1 in the prefix
<msk[m]>
shouldn't it just be /usr
micro_O has joined #kisslinux
<noocsharp>
yes, and DESTDIR="$1"
<noocsharp>
im rebuilding right now to see if this fixes it
<msk[m]>
same
<msk[m]>
I should have checked the build file before asking on here
<noocsharp>
well i'm the maintainer so it's my problem too
<msk[m]>
you maintain zathura-pdf-mupdf and libmupdf?
<noocsharp>
yes
mrlix has quit [Ping timeout: 246 seconds]
<msk[m]>
thanks for the packages, I've been using them for a while
<msk[m]>
I think it worked
<msk[m]>
no errors at the end
<msk[m]>
and `kiss manifest libmupdf` isn't just .cache/kiss
<msk[m]>
I'll try rebuilding zathura-pdf-mupdf now
<GalaxyNova>
make a github issue
<GalaxyNova>
oh nvm you fised it
<msk[m]>
everything works
<msk[m]>
great
<msk[m]>
thanks noocsharp
<noocsharp>
np, i'll send patches to fix the repo this evening
<msk[m]>
readelf still adds "+libressl" to libmupdf though, not sure if that's just me
<noocsharp>
not sure how this didn't cause issues earlier...
<noocsharp>
it does openssl for me msk[m]
<noocsharp>
i don't have libressl installed
monke has joined #kisslinux
<msk[m]>
I can't remove libressl, kiss stops me for that reason
micro_O has quit [Ping timeout: 246 seconds]
<msk[m]>
same thing with a few other packages
<sad_plan>
does it ever occur that -march=native is.. wrong? im just asking, as native for me is outputting a different result than what I belive I actually got. from what I can find, it outputs the version of cpus before mine.
<dilyn>
msk: use KISS_FORCE=1 to make kiss ignore dependencies, and then do kiss b $(kiss-revdepends libressl) to rebuild all of the packages that previously depended on libressl to link against openssl instead
<msk[m]>
what about packages that have libressl in the depends file?
<dilyn>
doesn't -march=native just select whatever gcc gets out of /proc/cpuinfo?
<msk[m]>
should I fork and replace them in depends
<dilyn>
msk: that's what kiss-revdepends libressl will find. it just greps /var/db/kiss/installed for libressl and gives a list of packages
<dilyn>
if you rebuild them it'll get fixed
<sad_plan>
dilyn: one would assume so. but in my case, iirc, it selecets zenv1, and I belive I got zenv2.
<dilyn>
unless you're still using the kiss-community/repo upstream, in which case you should switch...
<dilyn>
what's your cpu?
gtms has quit [Remote host closed the connection]
<sad_plan>
3700u
<dilyn>
a mobile chip?
<sad_plan>
yeha
<sad_plan>
yeah*
<dilyn>
iirc amd mobile chips lag behind one generation, intel's mobile cpu's are one ahead
<dilyn>
oh christ. they call it zen+, which came after zen1, but before zen2... wtf amd
<sad_plan>
lol wut
akira01 has quit [Ping timeout: 255 seconds]
<dilyn>
okay, wikipedia to the rescue. zen 2 is the successor to zen+; so gcc is probably correct
<sad_plan>
cool
micro_O has joined #kisslinux
<protonesso>
Hiii
<acheam>
hi protonesso , it's been a while
<protonesso>
Yup
GalaxyNova has quit [Quit: Whoooooshh]
micro_O has quit [Ping timeout: 255 seconds]
micro_O has joined #kisslinux
claudia02 has joined #kisslinux
claudia02 has quit [Client Quit]
micro_O has quit [Ping timeout: 255 seconds]
<micr0>
I started a kiss-testing project. Goal is to be able to smoke test kiss and package updates. Currently, it reliably errors out on the cache dir.