<dskuza>
yeah but who gives a shit about android jvulc
<jvulc>
I do.
<jdhartley>
nobody
<joelteon>
android users
<joelteon>
listening to my friends speak, android has apparently absolutely demolished iOS in the mobile sector
<joelteon>
nobody uses iphones anymore
<dskuza>
give me things to tell the dev i like yelling at him
<jvulc>
The thing I like about IRCcloud, which I can't find on a lot of other IRC clients: Working push notifications on Android.
<jvulc>
No need to keep the app running constantly in the background either for notifications like Andchat. It friggin' works + most of the time, it syncs the notifications from desktop webchat to mobile.
<niggler>
joelteon you are in a bubble
<joelteon>
yeah that's what i thought
<niggler>
iPhones are king in most places
<niggler>
starbucks, dunkin donuts
<jvulc>
I've seen a whole lot of Android devices though, particularly the S III, in the last year whether it be old or young.
* jvulc
switched from iOS to Android last year and haven't regretted it one bit.
<niggler>
I've played with quite a few android devices over the past month but every one of them pissed me off
<niggler>
itll be stupid things like inconsistent facial recognition
<jvulc>
Fragmentation be damned. They can't use that dialog anymore because I have three different devices on my desk from three different manufacturers and they all have the same OS ... run equally well, no app incompatibilities.
<jvulc>
I never use the face unlock.
<niggler>
inconsistent doesn't mean that the devices handle things differently
<niggler>
inconsistent here means that i need to replicate the lighting conditions and the exact facial expression to get a match
<jvulc>
I was just speaking in general.
<jvulc>
The face unlock is just a gimmick to show your friends.
<niggler>
but even apple's gimmicks "work"
<jvulc>
Anyone that actually wants to keep their phones locked will either use pattern or pin locking.
<niggler>
work meaning that they do what you expect
<jvulc>
I'm sure Siri wasn't 100% accurate when it was released. How's Passbook and Game Center?
<joelteon>
passbook is great
<joelteon>
game center is game center
<devyn>
good morning all
<devyn>
I am at the mall
<devyn>
capitalism for all
<joelteon>
cool
<joelteon>
i am testing hsvm
<joelteon>
it's slow work
<ELLIOTTCABLE>
prophile!
<ELLIOTTCABLE>
jvulc|znc: they absolutely *can* make that argument.
<ELLIOTTCABLE>
jvulc|znc: the problem with fragmentation isn't one that shows up, at first glance, to a user like you.
<ELLIOTTCABLE>
jvulc|znc: the way *you* will experience the fragmentation problems, is through a sub-par selection of apps.
<ELLIOTTCABLE>
Quite simply, nobody skilled or sane wants to develop for Android.
<ELLIOTTCABLE>
Quite beyond it being goddamn Java, it's *as bad as*, or worse than, dev'ing for the Web (huge inconsistencies, having to painstakingly test your work across a thousand target platforms, having to deal with users using obscure or weird platforms … etceteras … and even more reasons, that I don't want to get into here,)
<ELLIOTTCABLE>
but *unlike* the web, they A) have to *pay out the nose* for every single platform they'll have to test on, and B) they *have an alternative*.
<ELLIOTTCABLE>
With the web, it's “deal with the DOM and implementational inconsistencies, or walk the other way.”
<whitequark>
ELLIOTTCABLE: A) android is majority; B) it's not as bad as you make it look
* whitequark
shrugs
<purr>
¯\(º_o)/¯
<whitequark>
I mean, from the developer's POV.
<niggler>
whitequark 90% of the excess profits in any industry is generally in the top 10% of the market
<ELLIOTTCABLE>
On mobile, it's “deal with the OS and device fragmentation, and end up with an inferior product for *more work* that makes you feel bad about yourself,” or “just go build a superior app in a quarter of the time for iOS and feel good about yourself.”
<ELLIOTTCABLE>
Obviously, in many situations, the Android or iOS apps aren't *optional* (platform coverage, business concerns, etcetcetc),
<ELLIOTTCABLE>
but what that *really* means, in the real world, is that the skilled developers who won't take that shit, migrate back to the one they'll enjoy working on; and the newbies and interns and those who don't have the skills or wherewithal to control their own working environment end up slaved to doing the Android version …
<ELLIOTTCABLE>
… or, if the team's too small, it means that the developers who'd rather be putting that testing and annoyance time into more features and better UX, will half-arse the Android version, only putting in the effort they know they can reproduce across the fragmented install-base … and then put the *true* love and effort, their skill and obsession as
<ELLIOTTCABLE>
developers, into the iOS app.
<ELLIOTTCABLE>
and you end up with shit like twidroyd for Android, and true works of art like Tweetbot for iOS.
<ELLIOTTCABLE>
disclaimer: I haven't used Android in approximately a year. If something has hugely changed in the fragmentation space, and now all of the developers are *happy* to be trying to build mind-shatteringly-awesome UXes in Java for a thousand Android variants, that's awesome. More power to them. But last I've checked, it was a huge problem; one bad enough to kill
<ELLIOTTCABLE>
any interest I have in the platform.
<purr>
<alexgordon> ← the maths is strong in this one
<ELLIOTTCABLE>
also related: saw an exciting news article a couple of weeks ago, about Google taking a new approach to killing fragmentation in the Android space. Forget the specifics.
<niggler>
ELLIOTTCABLE google play framework
<niggler>
it is privileged and can be updated outside of the android OS cycle
<whitequark>
... which means that now there are exactly zero open-source mobile OSes which are *any* good.
<whitequark>
fuck.
<whitequark>
like, for real.
* whitequark
throws his fucking phone out of his fucking window.
<niggler>
why are you surprised
<niggler>
open source isn't a winning strategy
<niggler>
didn't ellison say that once?
<whitequark>
niggler: I'm not surprised, I'm annoyed
<whitequark>
besides, open source isn't a business model and has zero things to do with it
<whitequark>
it's a development model, and as a development model, it's far superior to *everything* else.
<niggler>
its funny how certain open source products actually ended up monetizing
<whitequark>
well, I think android was in it from the start, sort of
<whitequark>
selling consumer shit isn't exactly a place when you can be all white and fluffy and stallmanish
<whitequark>
(which is why I'm not surprised, at all)
<ELLIOTTCABLE>
whitequark: yep, trudat
<niggler>
they started that way to be opposite of apple
<niggler>
apple was closed etc etc
<niggler>
android was open
<whitequark>
ELLIOTTCABLE: which of the sentences?
<ELLIOTTCABLE>
I love the idea of webOS
<ELLIOTTCABLE>
re: you finally admitting that Android is a worthless pile of shit.
<niggler>
webOS was axed
<ELLIOTTCABLE>
(worthless, speaking as a consumer; worthless, speaking as a user; and worthless, speaking as a dev.)
<whitequark>
ELLIOTTCABLE: I don't, lol
<purr>
lol
<ELLIOTTCABLE>
webOS still exists!!!!! IN THE SHADOWS. SNEAKILY. EATING YOUR CORN-RINDS.
<whitequark>
right now I like where it is and how it works
<ELLIOTTCABLE>
but seriously. I really wish there was a Linux of mobile.
<whitequark>
it completely satisfies my needs and does it in the way I want.
<ELLIOTTCABLE>
Something truly open, something with a modicum of traction, and something that *wasn't* goddamn fragmented-ass Android.
<whitequark>
but that's not going to last, and *that* is what I hate.
* ELLIOTTCABLE
shrugs
<purr>
¯\(º_o)/¯
<ELLIOTTCABLE>
yes, and I can say the same thing about Apple.
<whitequark>
you can't have open and no fragmentaiton
<ELLIOTTCABLE>
but neither Android-servicing you, or Apple-servicing me, does anything to refute my statement:
<ELLIOTTCABLE>
that Android is a worthless pile of shit, as legitimate-satisfying-truly-open-platforms goes.
<whitequark>
there are 0 truly open platforms
<ELLIOTTCABLE>
and Apple's platform, although infinitely superior, obviously doesn't apply.
<whitequark>
PC isn't
<niggler>
whitequark the market has spoek
<niggler>
*spoken
eligrey has joined #elliottcable
<whitequark>
niggler: I don't care
<niggler>
and it doesn't give a shit about openness
<ELLIOTTCABLE>
niggler: “whitequark the market has spoek” wat.
<purr>
beep.
<whitequark>
about market
<whitequark>
at all.
<niggler>
textual lagging
<ELLIOTTCABLE>
bbl being a productive adult bye love you <3
<niggler>
whitequark just make your own
<niggler>
rambo that shit
<whitequark>
brb fucking with a proprietary self-bricking pile of shit of a JTAG adapter
<whitequark>
niggler: the farther away from the consumer, the more open things are
<niggler>
again, the market has spoken
* whitequark
sighs
<whitequark>
what you're saying essentially translates to "most people are morons"
<whitequark>
which is true
<whitequark>
and annoying
<whitequark>
and not surprising.
* whitequark
shrugs
<purr>
¯\(º_o)/¯
<niggler>
i used to fight
<niggler>
then i realized it was pointless
<niggler>
as an individual I don't have the power to change the landscape
<whitequark>
fighting doesn't work, of course. you can still work around that, if you have the motivation
<niggler>
i don't have the resources to build a compelling alternative
<whitequark>
there are people who do
<whitequark>
e.g. GTA04
<whitequark>
though that's not quite what I meant by working around
<joelteon>
my website dev environment takes 47 minutes to install from scratch
<whitequark>
foundry (without ocaml or llvm present) finishes the test cycle in a little under 6m
<whitequark>
and I consider that high
<joelteon>
well, i have to build and install yesod
<joelteon>
i really should RPM it
<joelteon>
then it'll only take like 5 min
<joelteon>
but i don't know how to rpms
<joelteon>
i'll find out
<whitequark>
ghc is really fucking slow
<whitequark>
also travis is deb
<joelteon>
oh right
<joelteon>
ghc isn't that slow
<joelteon>
yesod is huge
<joelteon>
some of that is compiling imagemagick, too
<whitequark>
what, 7944 lines of log O_o
<joelteon>
which takes a few minutes
<joelteon>
yeah, hang on a sec
<whitequark>
I already wanted to have nothing with haskell
<whitequark>
now I want to have even less
<joelteon>
my website has 178 dependencies
<whitequark>
yes exactly
<joelteon>
that's why it takes that long
<whitequark>
what the fuck
<whitequark>
even rails needs less
<joelteon>
most of that is yesod
<joelteon>
http stuff
<whitequark>
yes exactly
<whitequark>
EVEN RAILS NEEDS LESS
<joelteon>
that's baked into rails
<joelteon>
and ruby is an interpreted language
<whitequark>
so?
<joelteon>
bundle install in the outright repo takes 30-ish minutes
<whitequark>
I don't think so
<whitequark>
hang on a sec
<joelteon>
ok, but it does
<whitequark>
perhaps you're doing it wrong then
<joelteon>
ok
<whitequark>
so, our really huge app with just under 200 deps and database-hitting tests
<joelteon>
the resolver is pretty slow in low ram
<whitequark>
finishes the test cycle in 9min 37sec
<whitequark>
on travis
<joelteon>
what app is that
<whitequark>
amplifr.com
<joelteon>
huh
<joelteon>
well, if you put it that way
<whitequark>
though you won't see the admin interface anyway
<joelteon>
our pull request builds on jenkins
<joelteon>
the dependency installing bit takes about 60 seconds
<joelteon>
bundle's resolver is insanely slow in low ram
<joelteon>
but the test boxes have 16GB each
<whitequark>
evidently whichever ram travis has is not a problem
<joelteon>
yeah, so dropping and recreating the databases + migration
<joelteon>
takes about
<joelteon>
1:30
<whitequark>
though, hmm, if they provide less ram to opensource stuff that might be
<whitequark>
all my shit runs on their paid service
<joelteon>
it's the tests that takes awhile
<joelteon>
take*
<joelteon>
selenium is fairly slow
<joelteon>
anyway, there's no compile time in ruby, so an analogous test setup would be me setting up a package for my site that includes all the dependencies
<joelteon>
then just running yesod test
<joelteon>
which takes like 45 seconds
<whitequark>
well, look what I said above
<joelteon>
where
<joelteon>
you said a lot of stuff
<whitequark>
ah, you mean pre-compile the site
<whitequark>
well
<joelteon>
yeah, i mean package it, install it with apt during the test, and yesod test
<whitequark>
it's not really a fair comparison
<whitequark>
to begin with a buildbot is supposed to build
<joelteon>
yeah
<joelteon>
and in ruby, you don't build
<joelteon>
come to think of it, i might be able to install quite a bit of these dependencies with apt
<joelteon>
lens does that and the builds take 9 minutes
<whitequark>
what the fuck is with names in haskell
<whitequark>
it's even worse than in ruby
<joelteon>
ok
<joelteon>
huh, there's libghc-yesod-dev, i wonder what version it is
<whitequark>
regexps are regular, your grammars are context-free
<whitequark>
so you can't use nfa in general case
<whitequark>
the simplest thing to do is memoization
<whitequark>
this turns exponential time complexity into linear space
<whitequark>
complexity
<whitequark>
which in practice is still a mess
<joelteon>
ok
<whitequark>
because if you memoize every parser rule, your space's going to blow up
<joelteon>
what do I memoize?
<whitequark>
hm
<whitequark>
i dunno how to integrate it with your parser combinators
<whitequark>
i've seen a guy who also used parser combinators in haskell to solve this problem by memoizing only where needed, ie manually placed memoization
<whitequark>
but the article's in russian so I doubt it will help
<joelteon>
and as I expected, libghc-yesod-dev is like 3 versions old
<whitequark>
it's debian
<whitequark>
so of course
<joelteon>
yeah :\
<joelteon>
the only worth I could get out of backporting this to old yesods is faster tests
<joelteon>
that or I just build a yesod package myself
<whitequark>
it's rather trivial to build new debian packages
<whitequark>
well, new versions
<whitequark>
if you have an old one
<joelteon>
yeah i dunno how to do it but it can't be that hard