<< 06-07-2013 >>
01:03:39*BitPuffin joined #nimrod
01:11:51*DAddYE quit (Remote host closed the connection)
01:38:04*q66 quit (Quit: Leaving)
02:25:37*BitPuffin quit (Ping timeout: 256 seconds)
03:24:50reactormonklib/system.nim(1796, 23) Error: cannot open 'panicoverride'
03:24:52reactormonkhmm
03:25:36reactormonkAraq, forgot to check that in?
03:52:15*Associat0r joined #nimrod
03:52:15*Associat0r quit (Changing host)
03:52:15*Associat0r joined #nimrod
05:55:08*BitPuffin joined #nimrod
05:55:32*OrionPK quit (Read error: Connection reset by peer)
06:01:18*BitPuffin quit (Ping timeout: 264 seconds)
07:49:23*exhu joined #nimrod
07:50:05exhui regenerated csources and build/nimbase.h and compiler/mapping.txt have changed, is it normal? what are those files for?
08:13:11Araqreactormonk: nope, you need to provide it, look at the example in tests/manyloc/standalone
08:13:35Araqexhu: it's normal and mapping.txt is used by niminst to determine which C files exist
08:15:41exhui've filed a pull request
08:15:59exhuwithout those files
08:16:59exhudon't know why previous compilers built ok, because now on both machines they don't link.
08:17:23exhubut on mac it did link ok (clang)
08:19:24Araq what's "both machines"? windows?
08:46:37Araqbbl
09:01:05*q66 joined #nimrod
09:20:34exhuubuntu 13.04 64-bit gcc
11:11:13*exhu quit (Quit: Ex-Chat)
11:27:29*Trix[a]r_za is now known as Trixar_za
11:35:07*EXetoC joined #nimrod
12:21:01*BitPuffin joined #nimrod
12:22:34*Trixar_za is now known as Trix[a]r_za
12:56:45BitPuffinAraq: shouldn't the competetive concurrency model thing you're working on be added to the roadmap on the website?
13:15:53*gradha joined #nimrod
13:49:37*BitPuffin quit (Ping timeout: 276 seconds)
13:53:19*fowl quit (Ping timeout: 260 seconds)
14:05:45*fowl joined #nimrod
14:14:01EXetoCfowl: yo
14:15:41dom96yo yo
14:21:55EXetoClo
14:37:09gradhanihathrael: I believed yesterday to have replicated the bug you had but it turns out the problem was a stale/broken nimcache, after deleting the directory I could not reproduce it any more
14:39:22*apotheon quit (Ping timeout: 246 seconds)
14:40:52*apotheon joined #nimrod
14:40:52*apotheon quit (Changing host)
14:40:52*apotheon joined #nimrod
14:41:06gradhanihathrael: wait, I got it back, the problem is using suggest without a nimcache, could you try modifying your code to start the server and perform a "compile" command before querying for suggestions?
14:44:38gradhaalso I get to run a caas test succesfully once and it fails on the next run if I don't remove nimcache previously
14:57:19*zahary_____ joined #nimrod
14:57:19*zahary_____ is now known as zahary_
15:21:56*fowl quit (Quit: EliteBNC free bnc service - http://elitebnc.org)
15:36:27reactormonkAraq, kk
15:38:18*fowl joined #nimrod
15:49:03*BitPuffin joined #nimrod
16:10:36EXetoCBitPuffin: have you written any Nimrod code recently?
16:11:04BitPuffinEXetoC: I'm still plowing through the manual
16:11:12BitPuffinBut I'm starting a project on monday
16:11:48gradhamonday? sounds like serious stuff
16:11:54EXetoCok
16:13:21BitPuffingradha: why haha
16:13:56EXetoCmaybe he thinks you will get monies for it
16:13:56gradhajobs tend to start on mondays, are you doing something for work in nimrod?
16:15:36EXetoCI doubt it :p but that would be awesome
16:16:39BitPuffingradha: well it's not so much for work, but I'm gonna be more disciplined with working on my own projects and doing freelance stuff, so while I'm applying and waiting for responses etc, I'll work on my personal project that is serious
16:17:29gradhaI used to think that way (being disciplined) until I bought xcom for ios two weeks ago and have hardly made anything IRL
16:19:26BitPuffinhaha :D
16:43:11EXetoCI have no discipline whatsoever. I wonder if the lack of it is considered a medical condition
16:46:22EXetoChappy coding
16:46:26*EXetoC is now known as EXetoC_
16:47:10BitPuffinheh well me neither
16:47:18BitPuffinbut I'm attempting to aquire it
16:47:20BitPuffingonna get up earlier
16:47:23BitPuffinand just work
17:04:24*OrionPK joined #nimrod
17:24:30*BitPuffin quit (Ping timeout: 268 seconds)
18:06:45*apotheon quit (Ping timeout: 264 seconds)
18:07:20*apotheon joined #nimrod
18:07:20*apotheon quit (Changing host)
18:07:20*apotheon joined #nimrod
18:13:50*apotheon quit (Ping timeout: 268 seconds)
18:19:50*apotheon joined #nimrod
18:19:50*apotheon quit (Changing host)
18:19:50*apotheon joined #nimrod
18:24:09*EXetoC_ is now known as EXetoC
19:03:40reactormonkgradha, sweet stuff
19:13:01*alexandrus joined #nimrod
19:14:21*q66 quit (Remote host closed the connection)
19:14:50*q66 joined #nimrod
19:19:16gradhayou are eating something good?
19:19:41alexandrusmoin
19:20:11gradhahi alexandrus
19:20:15alexandrushow are you?
19:20:28gradhadebating on what to do next: play or play
19:20:29reactormonkgradha, the idetools
19:20:51alexandrushmm, one could also develop something xd
19:20:57gradhareactormonk: it's all just text nobody reads anyway
19:21:08gradhareactormonk: more impressive would be actually fixing idetools
19:22:03reactormonkgradha, I read that kind of text ;-)
19:22:13reactormonkgradha, got a bunch of failing tests?
19:22:53gradhamost of the tests fail for me, either in proc mode or caas
20:07:20Araqhmm tomorrow is Sunday ... maybe I should have bought some food ...
20:07:55alexandrusso tomorrow is a diet day
20:11:02EXetoCyes the best diet ever
20:11:11Araqmeh .. there are always gas stations
20:12:06*gradha quit (Quit: bbl, need to watch https://www.youtube.com/watch?v=1ZZC82dgJr8 again)
20:20:16EXetoCno food stores open on sundays?
20:20:39Araqnot in germany, no
20:20:46EXetoCoh man
20:21:33EXetoCalexandrus: eating is the main focus of proper diets you know :p
20:21:47alexandrusreally? xD
20:37:56EXetoCyeah, just avoid the bad stuff
20:38:57OrionPKglad I live in a secular country
20:39:08OrionPKwhere they just shut down liquor stores on sunday >:(
20:39:17OrionPK(also car dealerships)
20:45:29EXetoCI didn't know that the term 'secular' implied that as well :p
20:48:29*exhu joined #nimrod
20:49:08exhui feel like a seasoned tester here -)
20:50:26exhueither the compiler doesn't build or internal error or segmentation fault in generated program
20:57:45*exhu quit (Quit: Ex-Chat)
21:06:37Araqexhu, come back! perhaps I can help you
21:14:25*OrionPKM joined #nimrod
21:45:44reactormonkwhat's the borrow pragma?
21:46:05Araqit steals an implementation for a 'distinct' type
21:55:30*BitPuffin joined #nimrod
22:10:18*Sergio965 joined #nimrod
22:12:03AraqBitPuffin: the roadmap says "a shared memory garbage collected heap will be provided"
22:12:50BitPuffinAraq: and that's the competetive thingy?
22:13:31Araqno ... ;-)
22:13:54BitPuffinwell then what about it lol!
22:14:13Araqwell it's about concurrency, so there :P
22:14:43Araqalso the homepage is more about what we've accomplished rather than about features we hope to get some day
22:14:43BitPuffinhaha, you mean because the "shared memory" part hehe
22:19:35*OrionPKM quit (Remote host closed the connection)
22:31:48BitPuffinAraq: I'm trying to find some info about what competetive concurrency actually is, any recomendation?
22:32:02BitPuffincompetitive
22:33:10alexandrusone could imagine so many things x
22:34:07Araqwell I made up that term
22:34:12BitPuffinoh
22:34:19BitPuffinso that's why there is no papers about it
22:34:30EXetoCconcurrency that doesn't suck I imagine
22:34:42alexandrusconcurrency with race conditions xD
22:34:57alexandrusone process sabotages the other xD
22:35:03EXetoCrolf
22:35:13BitPuffinhörru rolf
22:35:15BitPuffinslutaa
22:35:17BitPuffinsluta rolf
22:35:21Araqfor me it means safety, no deadlocks no race conditions, no live locks
22:35:32alexandruslive locks?
22:35:51alexandruslike dead ends in a petri net?
22:36:30BitPuffinAraq: so what's the competitive about, can you even do that stuff without immutability stuff
22:38:07BitPuffinstuff.
22:40:02EXetoCstuff! STUFF!!
22:40:24EXetoCI need more stuff. where can I get some money?
22:42:58Araqimmutability means you can read from a data structure safely without any synchronization
22:43:22Araqso it's hard to argue it's pointless for concurrency
22:43:25BitPuffinyeah exactly
22:43:41BitPuffinbut how does your model solve the problems
22:43:47alexandruspersistence
22:44:23Araqthe problem with immutability is that its time dependent
22:44:24BitPuffinor rather, whet exactly is the model
22:44:36Araqand type systems suck at modelling time
22:44:49alexandruscould you elaborate that a bit?
22:45:09alexandrusthe time dependence here
22:46:02*zahary__ joined #nimrod
22:47:25Araqwell for a start you need to construct the data structure
22:47:51alexandrusthat so far is understood
22:48:04Araqand if it's not a tree, it's hard to construct it and keep it immutable during construction
22:48:07alexandrusso there is a time interval, where the data is created, at some point it is complete
22:48:37*zahary_ quit (Ping timeout: 248 seconds)
22:48:47alexandrusi guess i would follow Chris Okasaki on that (Purely Functional Data Structures)
22:48:59alexandruswhere each construction step creates a new persistent data structure
22:49:05Sergio965Immutability is in no sense time dependent.
22:49:11Sergio965If it is, it's not immutability.
22:49:13alexandruswith a lot of sharing
22:52:10BitPuffinyeah I'm not sure I see why it's time dependent
22:52:19BitPuffinit can't be used until it's constructed sure
22:52:28BitPuffinbut what's that got to do with concurrency
22:53:08BitPuffinif that was a problem then it would be a problem on one thread too
22:54:02alexandruslets say, we have some data structure over an index I
22:55:02alexandruswe could run some parallel algorithm on it, but we first have to distribute the necessary parts of the data structure to the nodes
22:55:16alexandrusthis step could benefit from persistence
22:55:34alexandrussince the data structure won't break during that
22:55:55Sergio965What does concurrency have to do with persistence?
22:57:02BitPuffinwell not necessarily anything, but concurrency benefits from persistence
22:57:12alexandrusits just my 5 cents
22:57:28alexandrusthat the process of distributing data to a concurrent nodes
22:57:31alexandrusmay benefit here
22:57:46alexandrusmaybe for real world scenarious one needs a kind of snapshot mechanism
22:58:26alexandruslike a distributed transaction, if we add the reverse
22:58:58Sergio965How does concurrency benefit from persistence? I don't understand how concurrency and persistence conflate in a positive way.
22:59:03alexandrusbut what do you think about that, Araq?
22:59:27alexandrusmaybe the wording is not right?
22:59:45alexandrusi took persistence from a book
23:00:01alexandrusi think its used differently in other contexts
23:00:29alexandrusjust, create the data once, cannot modify it ever again
23:00:41Sergio965Persistence = to persist = write to disk (usually).
23:00:45alexandrusimmutable, ~ java final
23:00:57Sergio965Java's final doesn't imply immutability all the time.
23:01:24alexandrusok...won't argue that, i am a java newbe
23:01:26Sergio965Or even half the time.
23:01:32Sergio965Just rarely.
23:01:39alexandrusmy languages are Ada, C++
23:01:50Sergio965Immutability is like doing const in C++.
23:02:05alexandrusif there is no alias and no const-cast involved
23:02:16Sergio965Well, no.
23:02:38Araqalexandrus: since you all know better than me, I'm quiet now
23:02:48alexandrusAraq?....
23:03:03alexandrusi wasn't about to quit you down at all...
23:03:12alexandrusplease, stay with us:-)
23:03:23alexandrusdon't be thrown of by my 5 cents
23:03:37Sergio965Araq: I challenged your assertion that immutability is time-dependent. I don't presume to know more than you. Simply, it seems to me, by definition, immutability has nothing to do with time. If an object is said to be immutable, it cannot be changed.
23:04:18Araqbreak construction apart into multiple steps and nothing can be immutable
23:04:49Sergio965What you're describing has all to do with atomicity and nothing to do with immutability.
23:05:16Sergio965The "atoms" are said to be immutability, a structure made of multiple atoms cannot, unless made atomic, be inherently immutable.
23:05:20Sergio965immutable*
23:05:27alexandrusAraq, are we speaking of functional construction or imperative construction?
23:05:35BitPuffinimmuntability is like doing let in nimrod :D
23:05:45alexandrusatoms are mutable, says the physicist xD
23:06:14alexandrusi have to take the time to learn nimrod first, currently perparing for a job interview...not so easy to do everything at the same time
23:06:23alexandrus(and prepare a talk about multi threading next week too)
23:06:47alexandrusmercy:-)
23:14:29AraqSergio965: that's way to philosophical
23:15:00Araqyou construct a data structure, then it doesn't change anymore and so you can read without synchronization
23:15:30Sergio965Araq: As long as the construction occurs atomically, then the entire data structure can be said to be immutable.
23:15:32alexandrusbut where is the time dependence "problem"
23:16:11EXetoCtime? sequence?
23:17:29Sergio965What is "time"? Where is "forward"? What is?
23:18:07alexandruswe don't want to go into physics here, do we? xD
23:18:08AraqSergio965: the construction often can't occur atomically
23:18:51Araqfor example, try to construct a graph with a cycle in a functional language
23:19:02alexandrusAraq, just wondering, if the construction is not atomically, can it be separated into construction of intermediate steps, where each is created atomically in your language?
23:19:51alexandrusi guess, thats only possible as I -> Node , IxI -> Bool
23:19:55Sergio965Araq: Self-reference doesn't imply that atomicity is impossible. (In Haskell, for example, with lazy loading, I can do x = 1 : x)
23:20:12Sergio965Now, I have an infinite, immutable cycle.
23:20:39AraqI know but laziness is cheating :P
23:20:48alexandrusähm, the term self reference is a bit....vague to me here
23:21:18alexandruspointer self reference vs recursion
23:21:24Sergio965alexandrus: It just means that the structure somehow refers to itself. Like a linked list that starts with some element X and somewhere in the middle, goes back to that element.
23:21:49Sergio965alexandrus: I would argue that recursion is inherently self-referential, but in a different sense. I was speaking purely about data structures.
23:21:49alexandrusok, so pointer self reference, but thats not x=1:x, which is just recursive
23:22:09Sergio965What do you mean? Of course it is.
23:22:34Sergio965If you iterate through x, you won't do it recursively, will you?
23:22:36alexandrusany pointers in here?...where?
23:22:44Sergio965Implicitly. That was Haskell.
23:22:51alexandrusi know haskell
23:22:59Sergio965Okay...
23:23:04alexandrusok, thats implementation detail
23:23:11alexandrusinternal...there are lots of pointers of course
23:23:23Sergio965The point is if you iterate through x, you'll get an infinite number of 1s.
23:23:26Sergio965Do you agree with that?
23:23:41alexandrussure, its an infinite list of 1:(1:(1:(1:....))))
23:24:35Sergio965Yeah, and it's infinite because it refers back to itself.
23:24:37alexandrusok, lets drop semantic stuff, for the abstraction haskell provides, thats not based on pointers
23:24:43alexandruseven if internally, its implemented like that
23:24:46Sergio965:\
23:24:56Sergio965You don't need the concept of pointers to be self-referential.
23:25:27alexandrus:-)
23:26:11alexandrusok, so now lets get to graphs with cycles in them...
23:26:30Sergio965Anyway, my point was that a self-referring data-structure can be created atomically. And, if all of its elements are immutable, the data-structure is immutable. Because of atomicity, there is no time-dependence..
23:26:45alexandrusi think recursive definition like with trees is not possible here
23:27:09Sergio965alexandrus: x = 1 : x is a graph with a cycle in it. Specifically, it's a graph with one node and one edge that goes from that node to itself.
23:27:12AraqSergio965: you don't need haskell for that, you can subsitute pointers by indexes to get the same
23:27:32Sergio965Araq: Agreed. It was just the easiest and most illustrative example I could think of.
23:30:49alexandrusok, next question...
23:30:54alexandrusthree nodes
23:30:55alexandrus1,2,3
23:31:27alexandrusnow, we have a directed edges (1,2), (1,3), (2,1), (3,1)
23:31:34EXetoCone went to the bathroom
23:31:43Sergio965Lmao.
23:33:37Sergio965alexandrus: Uh huh...
23:33:53alexandrusyes, i am trying a solution, but i don't get it quite right in your terms
23:33:59alexandrustwo cycles
23:34:11alexandrusa=1:2:a, b=1:3:b
23:34:14alexandrusbut how to combine them
23:34:21Sergio965Araq: What is exactly does the effect system let me do? I think I understand the gist of it from the manual (track exceptions, tag them, contain them, etc), but how would I use it in practice?
23:34:24alexandrusits a valid graph, thats obvious
23:35:27Sergio965alexandrus: What are you trying to do?
23:35:31alexandrusx=1:2:1:(2,3) there is some kind of choice in here...which defies the style you used
23:35:41alexandrusi don't get that graph in there
23:35:53alexandrustrying to get beyond the x=1:x
23:36:28AraqSergio965: there is still the ~ missing in the language unfortunately; proc main() {.tags: [~FDb].} # compiler ensures main does no database access
23:36:42Sergio965Araq: That's neat.
23:36:56Sergio965Araq: Statically, I assume?
23:37:12Araqyep
23:37:39Sergio965Do I have to supply the .raises: [ blah ] pragma to all the function that can possible raise FDb or is it inferred?
23:37:55AraqFDB is no exception here, it's a tag
23:38:02Sergio965Oh I see.
23:38:25Sergio965In that case, do I have to tag all the procs with FDb?
23:38:33Araqit's all inferred as far as possible
23:38:59Araqand the stdlib already uses the FDb tag
23:40:24Sergio965I guess what I'm wondering is how you get through something like this: proc sometimes_db(yes: bool) = if (yes) db_access() else something_else();. Will a proc that call sometimes_db(complicate_false()) with the tag ~FDb be allowed?
23:40:46Araqno the tracking is flow insensitive
23:40:54Sergio965Ah I see.
23:41:08Sergio965That would have been the question to ask. Ha.
23:42:10Araqnowadays we got a flow analyzer but I still think it's a bad idea to make it flow sensitive
23:45:47BitPuffinoh that reminds me, Araq, is there any lazyness in nimrod?
23:46:20AraqBitPuffin: we have 'iterator' and first class functions
23:46:52Araqalso the macro system enables some form of laziness
23:47:50alexandrus<- admits, that he has a issues again...
23:48:06alexandruswasn't lazyness not: evaluate only when needed?
23:49:04BitPuffinAraq: hmm, how does macro system enable that?
23:50:03AraqBitPuffin: template iff(a, b, c: expr): expr = if a: b else: c
23:50:29Araqput it into a proc instead and see the laziness go out of the window
23:51:43Araqok "enable" is perhaps a bit strong but it at least it doesn't lose laziness
23:51:52alexandrusok...
23:51:58alexandrushave a good night:-)
23:52:02*alexandrus quit ()
23:53:54BitPuffinhmm