<dsamarin>
whitequark: if you look at the citations, sqlite improved speed in spkg
<dsamarin>
dpkg*
<dsamarin>
plus there are other alternatives
<whitequark>
I also know that it used to be a huge problem in firefox
<whitequark>
well
<dsamarin>
especially the way it's done now is terribad
<whitequark>
my point is, sqlite is good (why not), but you're gonna be stuck fsyncing around, just as every other package management system
<dsamarin>
so there's absolutely no way around it
* whitequark
nods
<dsamarin>
at least pacman is pretty speedy already
<dsamarin>
so that's a nonissue
<whitequark>
what would be the killer feature?
<whitequark>
ok
<whitequark>
you know, which makes me migrate from dpkg
<dsamarin>
heh i doubt you would use a system not supported by your distro anyway, but i'm still brainstorming what the killer feature should be
<dsamarin>
another person was saying the features were too boring
<whitequark>
it's something *expected* in this day and age
<whitequark>
so it's not really features
<whitequark>
more like, absence of any of those would be busg.
<whitequark>
*bugs.
<dsamarin>
i think package managements systems now are mostly out-of-your-way now so that's a good question
<whitequark>
exactly
<dsamarin>
what about support for installing packages in virtual envs/chroots/containers?
<whitequark>
what's wrong with debootstrap?
<dsamarin>
what is wrong with it
<dsamarin>
whitequark: android has the only package manager i know that uses binary diffs for updates. in google chrome, updates with courgette are .7% the size of a full update (what apt, rpm does now)
<whitequark>
that's definitely cool
<dsamarin>
that's a 0.7 percent
<whitequark>
but why not patch apt?
<dsamarin>
hells fricken yeah
<whitequark>
everyone would be grateful as fuck
<whitequark>
I'll reply: because mirrors don't have the capacity to build binary updates for everyone, for 22k packages in deb.
<dsamarin>
i'd rather start from scratch because of some security vulnerabilities apt has which if fixed would not be backwards-compatible
<whitequark>
such as?
<whitequark>
and how are you going to guarantee that you won't add completely new and unknown vulns?
<dsamarin>
it's never possible to guarantee but it's possible to take an existing infrastructure and fix known vulns
<dsamarin>
whitequark: it's possible to make something stronger with specific security design goals in place
<dsamarin>
i mean package management is a somewhat simple concept but people forget some basic things like signing metadata
<whitequark>
so it's just the reply attack
<whitequark>
it's not practically applicable
<whitequark>
for several reasons, the ultimate of which is that security.debian.org points specifically to security.debian.org
<whitequark>
maintained by debian.org and *without* any mirrors
<whitequark>
specifically to avoid this attack.
<whitequark>
and, as they helpfully mentioned, once upgraded the package manager won't install the old version on its own.
<whitequark>
i don't get that page, it's completely conceptual
<whitequark>
no proofs, specific listings, CVE numbers or anything
<whitequark>
bullshit
<dsamarin>
they managed to own their own mirror
<whitequark>
so what
<dsamarin>
apparantly
<whitequark>
I've explained above that it doesn't matter a slightest
<whitequark>
well, for deb
<dsamarin>
i know canonical has commercial software for multi-workspace package management
<dsamarin>
how about an open version?
<whitequark>
now that sounds like something really interesting.
<whitequark>
would you use, say, dpkg as a backend?
<whitequark>
you could be able to reuse packagekit or something.
<whitequark>
that'll give you deb rpm and maybe something else.
<whitequark>
see, no one really needs a *replacement* for dpkg for multi-workspace pkg management; they work on different levels.
<whitequark>
even apt is lower-level.
<whitequark>
canonical's thing is above apt too, I'm sure.
<dsamarin>
i'd rather make a new pkg system so that it's easier to implement diff updates
<whitequark>
"hey, we don't like how the buttons are rounded in Qt" "Suuuuure, let me just rewrite Linux"
<dsamarin>
plus ΩF:∅ has the resources to fund security audits, something other people don't do
<whitequark>
(though that is an adequate restriction of most open-source :p)
<dsamarin>
whitequark: it would be difficult to extend a current pkg system with those features
<whitequark>
does oftn have the resources to completely replicate dpkg and apt?
<whitequark>
since that's what you would need *before* doing that multi-workspace thing.
<dsamarin>
that's the plan
* whitequark
shrugs
<purr>
¯\(º_o)/¯
<dsamarin>
i mean i would love to be able to incorporate all my requested features into dpkg so much
<dsamarin>
seriously!
<whitequark>
try harder :p
<whitequark>
fork dpkg.
<whitequark>
still better than completely new system.
<whitequark>
not necessarily dpkg, fork anything existing.
<dsamarin>
maybe pacman
<whitequark>
maybe
<dsamarin>
fun fact: apt was the first widely known dependency resolution tool
<dsamarin>
i guess before you had to install the dependencies manually
<dsamarin>
even if it was packaged
<whitequark>
dsamarin: don't think so
<whitequark>
`make' is most widely known one :p
<dsamarin>
whitequark: omg wikipedia is wrong!
<whitequark>
doesn't freebsd's portage still use it?
<dsamarin>
heh
<dsamarin>
yeah
<jesusabdullah>
dsamarin: yeah slack's package system was like that
<dsamarin>
you mean slack's "package system" :)
<jesusabdullah>
yes
<whitequark>
you "mean" "slack"'s "package" "system"
* whitequark
"performs" "air quotes"
<dsamarin>
yes
<dsamarin>
"yes"
<jesusabdullah>
"\"air\" \"quotes\""
<dsamarin>
"oh \"god \\\"nesting\\\"\""
<dsamarin>
whitequark: how do you feel about packagers which allow multiple versions of the same package to be installed simultanously? (most commonly done with symlinking to the package version's dir)
<dsamarin>
e.g. nix
<whitequark>
fuck that
<whitequark>
I've had enough of it with rubygems
<whitequark>
it is an extremely bad idea
<dsamarin>
heh npm as well
<whitequark>
npm is different
<whitequark>
it has namespacing
<whitequark>
unix doesn't really
<dsamarin>
hm
<whitequark>
neither namespacing nor a working module system
<dsamarin>
right
<whitequark>
well, unix, or rubygems alike
<whitequark>
dsamarin: wait why are you in #ocaml
<dsamarin>
whitequark: studying sks
<whitequark>
sks which?
<dsamarin>
you know the synchronizing keyserver?
<dsamarin>
it's written in ocaml if you can believe it
<whitequark>
oh cool
<whitequark>
write your pkg manager in ocaml
<whitequark>
<3 ocaml
<purr>
Let it be known that whitequark hearts ocaml.
<dsamarin>
lol
<purr>
lol
<whitequark>
easily deployed, fast, etc
<whitequark>
well if not
<whitequark>
DON'T USE JS OR PYTHON
<whitequark>
or i will find and strangle you
<dsamarin>
oh hell no
<dsamarin>
i'd stuff a sock in my mouth if i did that
* dsamarin
is javascripted out
<whitequark>
why are you fucked out
eligrey has quit [Quit: Leaving]
<ec>
micahjohnston: your class-list makes me want to kill myself.
<ec>
“” physics, calculus III, CS (discrete structures, set theory type stuff), and beginning linguistics
<ec>
omfg. you're too smart. saddestface.
<ec>
damnit guys
<ec>
why do you have interesting conversations at times when I'm going to not want to read computer-shit the next time I log-in!
alexgordon has joined #elliottcable
cloudhead has joined #elliottcable
cloudhead has quit [Quit: Lost terminal]
sharkbot has quit [Remote host closed the connection]
sharkbot has joined #elliottcable
cloudhead has joined #elliottcable
Sgeo has quit [Read error: Connection reset by peer]
yrashk has quit [Ping timeout: 245 seconds]
<jesusabdullah>
lyngwistixxx
<jesusabdullah>
I don't *have* college classes anymore!
<jesusabdullah>
:)
<joelteon>
**
eligrey has joined #elliottcable
PragCypher has joined #elliottcable
cloudhead has quit [Ping timeout: 245 seconds]
<jesusabdullah>
shit
<jesusabdullah>
how do you print screen in osx again? :C
<vil>
Cmd+Shift+3
<vil>
or 4 to grab just a region
<vil>
4+Space to grab a window
<jesusabdullah>
okay
<vil>
shots will appear on the desktop
<joelteon>
I hate when fucking bugs disappear when you enter the DEBUGGER
<joelteon>
GOD DAMN IT
<vil>
lol debuggers
<purr>
lol
<joelteon>
DEBUGGER should not mean BUG DISAPPEARER
<vil>
real programmers debug with print statements while drinking unicorn tears
<jesusabdullah>
^^^^^^
<joelteon>
i don't HAVE any unicorn tears
<vil>
they just materialize when you become a real programmer
<vil>
sort of a catch-22
<joelteon>
darn
<vil>
the only other option is to kill a real programmer and take theirs
<joelteon>
ok
<joelteon>
are there any of them here
<vil>
dunno
<vil>
alexgordon: do you have any unicorn tears?
<alexgordon>
vil: I did tear a unicorn once, I but I think it healed
<joelteon>
you but you think it healed
<alexgordon>
me but me yes
<joelteon>
i think i'm going to give myself a hitler youth haircut
<joelteon>
and see if it works
<joelteon>
so I found out about a structure called a seminearring
<joelteon>
"semi-near ring"
<joelteon>
but the first time I read it I thought it was "semen earring"
<vil>
close enough
<joelteon>
and it was cool because you can write out the letters from 1 to 100 billion and find the middle letter using a binary search
<joelteon>
I still don't have the faintest fucking idea how they did that
PragCypher has quit [Quit: Leaving]
<jesusabdullah>
that definitely says semin earring
alexgordon has quit [Quit: Computer has gone to sleep.]
alexgordon has joined #elliottcable
alexgordon has quit [Remote host closed the connection]