Koterpillar 2017-01-26 13:45:08
well, if my door works today, it might break tomorrow
monochrom 2017-01-26 13:45:25
But Google's search algorithm is also "software", and it suffers just the same uncertainties as my brain. In fact I think Google is more complex than my brain.
orzo 2017-01-26 13:45:30
askhask: if you reproduce the exact circumstance for what it worked on, then i expect you'll see a similar result tomorow
askhask 2017-01-26 13:45:40
monochrom: really? i had read that airplane control was hugely heuristic
monochrom 2017-01-26 13:46:05
Oh, it's all relative isn't it?
monochrom 2017-01-26 13:46:39
Airplane control is more complex than my toy programs. But still laughably simpler than Google or high-frequency trading or ...
monochrom 2017-01-26 13:48:07
But you should look up David Parnas and his work on using formal methods for some airplane software. It shows you how this piece is still within reach for full analysis.
monochrom 2017-01-26 13:49:20
At which point, the question is no longer "will the software do the right thing tomorrow?", but rather "will the hardware sensors give true readings to the software tomorrow?"
geekosaur 2017-01-26 13:51:17
how can you tell anyone's brain works? maybe we're all figments of an AI pretending to be an IRC server
ggVGc 2017-01-26 13:51:41
what's the difference?
monochrom 2017-01-26 13:51:53
In that case you can still ask "does this figment work?" :)
Tuplanolla 2017-01-26 13:52:24
It would be pretty cool if you were all machines.
monochrom 2017-01-26 13:52:33
"What does 'my brain' mean?" --- A question on referential transparency.
monochrom 2017-01-26 13:53:25
I need to close this by sharing you my recent invention:
M0000 2017-01-26 13:53:28
question about a music application here: http://lpaste.net/351685
monochrom 2017-01-26 13:53:34
@quote monochrom lisp.*haskell
lambdabot 2017-01-26 13:53:34
No quotes match.
ggVGc 2017-01-26 13:53:38
What colour is your imagination? Referentially transparent
monochrom 2017-01-26 13:53:47
hrm, no? nevermind.
ggVGc 2017-01-26 13:54:13
monochrom: what's the question?
ggVGc 2017-01-26 13:54:25
sorry
ggVGc 2017-01-26 13:54:27
M0000:
monochrom 2017-01-26 13:54:30
(Something about "Every sufficiently advanced Lisp program contains a partial implementation of Haskell")
ggVGc 2017-01-26 13:54:39
you want feedback of your data structures?
ggVGc 2017-01-26 13:55:17
M0000: imo, the way you find out if your data structures are sound, is if the functions you need to write come out okay without needing to transform the data a lot
monochrom 2017-01-26 13:55:34
My advice would be "eradicate all type synonyms".
ggVGc 2017-01-26 13:55:52
sounds violent
dfeuer 2017-01-26 13:56:14
Is/was there a known problem with cabal/quickcheck/test-framework/whatever causing tests invoked via `cabal test` to freeze in a `pause` system call?
dfeuer 2017-01-26 13:56:36
Because I'm seeing that with cabal test seq-properties in containers....
M0000 2017-01-26 13:56:44
ggVGc: I have a "use case" and wondered what a Haskelly way to approach it is — although maybe it needs more work to determine
M0000 2017-01-26 13:57:32
it's big hierarchical data and the desire is write functions that can access pieces of it as well as look elsewhere in it — move "up" and "down" the hiearchy
M0000 2017-01-26 13:57:47
and modify it, too
M0000 2017-01-26 13:58:36
when you say "not need to transform the data", what I'm thinking is that certain data is fixed throughout the application, and certain data is computed… and I want to keep them separate, perhaps?
M0000 2017-01-26 13:59:17
Like if I have a hiearchy of fixed data, and then need to compute attributes of pieces of that hiearchcy, I store those attribtues in a separate data struct that identifies their position in the hiearchy?
dfeuer 2017-01-26 14:04:10
Oh, I figured it out.
dfeuer 2017-01-26 14:04:13
Never mind.
dfeuer 2017-01-26 14:04:22
Grey hole rather than black hole.
dfeuer 2017-01-26 14:04:24
*sigh*
monochrom 2017-01-26 14:04:36
Intrigung
orzo 2017-01-26 14:17:17
"A retainer is either the system stack, an unevaluated closure (thunk), or an explicitly mutable object." I tend to think I see only unevaluated closures in my heap graph. I don't really understand the otehr two cases
orzo 2017-01-26 14:17:48
can i tell what case i'm looking at by how it is marked in .prof ?
orzo 2017-01-26 14:18:34
is an explicitly mutable object something like an IORef or TVar?
monochrom 2017-01-26 14:19:27
The system stack holds actual parameters to functions. They can be the roots of all the heap objects you want live.
orzo 2017-01-26 14:22:26
in .prof, each of the "SET nnn = ..." is a reference to a single retainer object, right?
orzo 2017-01-26 14:23:52
and theres no way to tell by it whether it is a function parameter, unevaluated thunk, or mutable object, right?
codedmart 2017-01-26 14:24:18
If I have a list ["some.element", "element", "other", "some", "other.element"]. Is there a fn already for that or do I need to write my own?
codedmart 2017-01-26 14:24:43
If I want to see if any elements contain "element".
codedmart 2017-01-26 14:24:56
elem only matches exactly.
monochrom 2017-01-26 14:25:33
You should probably use "filter" with "isInfixOf" and then check for non-emptiness.
monochrom 2017-01-26 14:25:52
No, use "any" with "isInfixOf"
codedmart 2017-01-26 14:26:02
OK I was thinking of something like that but didn't want to if there was something already.
codedmart 2017-01-26 14:26:03
Thanks
kadoban 2017-01-26 14:31:47
Wonder when there's gonna be a stackage LTS with 8.0.2
ertes 2017-01-26 14:31:57
in (bracket o c k), are o and c masked by default, or do i need (mask $ \unmask -> bracket o c (unmask . k)) explicitly?
monochrom 2017-01-26 14:34:12
I forgot which one is c and which one is k.
xpika 2017-01-26 14:34:16
How can I programatically lookup the package for a particular module. Say map Data.Text to the package "text"
geekosaur 2017-01-26 14:34:48
xpika, hayoo is good for that. or if it's an installed package, there's ghc-pkg find-moduke
geekosaur 2017-01-26 14:34:51
*find-module
monochrom 2017-01-26 14:35:02
o and c are under mask.
monochrom 2017-01-26 14:36:04
k is under a restore.
monochrom 2017-01-26 14:36:13
err, unmask
xpika 2017-01-26 14:37:52
is there a stackage package for installing ALL of lts ?
ertes 2017-01-26 14:38:11
monochrom: thanks
ertes 2017-01-26 14:38:19
this should probably be documented
xpika 2017-01-26 14:38:29
geekosaur: thanks. Didn't know about ghc-pkg find-module
monochrom 2017-01-26 14:38:41
Yeah, would be nice to say it in the blurb so I didn't have to check source code.
orzo 2017-01-26 14:39:25
"This code guarantees that acquire is paired with release, by masking asynchronous exceptions for the critical parts. (Rather than write this code yourself, it would be better to use bracket which abstracts the general pattern)."
orzo 2017-01-26 14:39:40
that's from the blurb attached to Control.Exception.mask
monochrom 2017-01-26 14:39:48
Also, o and c are under the same mask scope.
orzo 2017-01-26 14:40:23
monochrom: that's evident from the code quote immediately above what i quoted
orzo 2017-01-26 14:40:33
so it is documented
monochrom 2017-01-26 14:40:39
orzo, we are referring to the doc of bracket
orzo 2017-01-26 14:40:40
maybe just not on 'bracket'
monochrom 2017-01-26 14:41:21
You will find yourself complaining at a similar phenomenon the next time you read a math book.
orzo 2017-01-26 14:42:50
i wish other languages would link source the way haskell/haddock do
monochrom 2017-01-26 14:43:02
A proof on page 439 would go like "Since f(x,y) = f(y,x), therefore...", and you will ask the teacher how do we know f(x,y) = f(y,x) in the first place.
monochrom 2017-01-26 14:43:20
And the teacher will answer "but it's from a theorem 200 pages ago!"
orzo 2017-01-26 14:43:27
heh
