<< 16-04-2014 >>

---

00:01:14	Varriount	Araq: My getFileInfo addition is almost done. I just need to add support for birthTime, check that the _get_osfhandle native call is available on all compilers, and finish up the unit tests.
00:03:51	*	askatasuna joined #nimrod
00:06:26	reactormonk	flaviu, the model is 744MB
00:08:22	flaviu	I guess the PStream would work best, I doubt you can make it fit the json in memory
00:25:29	reactormonk	did work best
00:25:51	flaviu	How big was the output?
00:28:20	reactormonk	the 750m
00:29:23	reactormonk	mostly zero, let's see if there's an error somewhere...
00:38:15	*	DAddYE_ joined #nimrod
00:39:06	*	DAddYE__ joined #nimrod
00:40:15	*	DAddYE__ quit (Remote host closed the connection)
00:41:45	*	DAddYE quit (Ping timeout: 258 seconds)
00:42:37	*	DAddYE_ quit (Ping timeout: 250 seconds)
00:58:26	*	nande joined #nimrod
01:14:25	*	q66 quit (Quit: Leaving)
01:19:36	*	brson quit (Ping timeout: 265 seconds)
01:21:17	*	brson joined #nimrod
01:24:04	BitPuffin	Skrylar: does not really apply in this case though but I see your point
01:35:59	*	mwpher joined #nimrod
01:36:12	reactormonk	I think the scala version might still be faster, but drinks memory as fuck.
01:37:47	flaviu	reactormonk: Enable profiling and see what you get
01:39:53	*	mwpher_ joined #nimrod
01:41:17	*	DAddYE joined #nimrod
01:41:50	*	mwpher quit (Ping timeout: 265 seconds)
01:41:50	*	mwpher_ is now known as mwpher
01:43:44	reactormonk	flaviu, nimrod profiling?
01:43:47	reactormonk	I missed that one.
01:44:01	flaviu	Yep, pretty neat: http://nimrod-lang.org/estp.html
01:44:16	BitPuffin	brb
01:44:40	*	mwpher quit (Client Quit)
01:45:27	EXetoC	waowaowowoa
01:45:46	*	DAddYE quit (Ping timeout: 258 seconds)
01:48:47	*	BitPuffin quit (Ping timeout: 252 seconds)
01:50:27	*	psquid joined #nimrod
01:51:40	*	Guest6461 quit (Ping timeout: 240 seconds)
02:14:51	reactormonk	flaviu, http://sprunge.us/RKOY hmmm
02:17:43	flaviu	assign.nim:129 var src = src # ugly, but I like to stress the parser sometimes :-)
02:19:03	*	DAddYE joined #nimrod
02:29:39	*	OrionPK quit (Ping timeout: 250 seconds)
02:33:33	fowl	flaviu, that not ugly, sometimes you need a local copy
02:34:00	*	Demos joined #nimrod
02:35:17	Demos	ugh why do hardware configuration programs suck so much
02:35:31	Demos	makes xorg.conf look like the best UX ever
02:37:40	*	Jesin joined #nimrod
02:38:36	*	Demos quit (Read error: Connection reset by peer)
02:39:20	*	Demos joined #nimrod
02:41:23	flaviu	fowl: Not my code, but its a two line method where the second line just delegates to another method that doesn't need a var
02:41:41	*	Jesin quit (Client Quit)
02:43:23	flaviu	Never mind, I see. It calls addr on src, so it needs that copy
02:46:07	*	Jesin joined #nimrod
02:49:22	reactormonk	can I iterate over the values of a table and modify it the same time?
02:49:30	reactormonk	or is that undefined behaviour.
02:50:21	fowl	reactormonk, mitems()
02:50:39	*	BitPuffin joined #nimrod
02:50:40	reactormonk	fowl, I want to delete some of them
02:51:13	reactormonk	*some of the items
02:53:07	fowl	o
02:53:28	fowl	make a list of keys to delete and do it after iteration
02:54:11	reactormonk	hmm, that's gonna be big
02:55:23	reactormonk	uhm, do we have a sortBy somewhere?
02:55:55	fowl	algorithm module
02:56:02	fowl	sort()
02:57:09	reactormonk	I'm not exactly sure my commit permissions correlate with my nimrod knowledge
02:58:07	fowl	you are the prime minister of git
02:58:21	fowl	dont worry about it
02:59:03	reactormonk	BOW TO ME
03:04:02	*	BitPuffin rises
03:06:33	*	nande quit (Ping timeout: 245 seconds)
03:08:13	reactormonk	relevantContexts = accumulateSeq(cc.pairs).sort(proc (a, b: tuple[string, int]) = cmp(b[1],a[1]))
03:08:17	reactormonk	^ find the syntax error
03:10:09	fowl	lol
03:10:41	fowl	reactormonk, i think the proc you send to sort has to return -1 0 or 1
03:11:18	fowl	other than that i dont see any error
03:12:12	fowl	although accumulateSeq probably need to be assigned to a var
03:13:05	fowl	reactormonk, sort doesnt return a seq, its for sorting in place
03:15:20	Skrylar	meep
03:15:33	Skrylar	i have a pet peeve about methods named "sort"
03:15:44	Skrylar	there are at least a dozen types of sort
03:17:04	reactormonk	fowl, oh, int?
03:17:05	flaviu	I hate compare functions with +1,0,-1. Why not just use an enum, which has 0 overhead and is 100x clearer
03:17:20	reactormonk	fowl, and I assume cmp does exactly that?
03:17:32	Skrylar	well, in nimrod the compare function is a boolean
03:17:42	Skrylar	IIRC x > y is templated back to x < y
03:17:46	Skrylar	* y < x
03:18:28	flaviu	> fowl: reactormonk, i think the proc you send to sort has to return -1 0 or 1
03:19:29	flaviu	So someone, somewhere, has decided that they prefer the weird version
03:19:47	reactormonk	flaviu, true false and nil otherwise?
03:20:09	flaviu	I don't know, sorry.
03:21:03	reactormonk	relevantContexts = targets.sort(proc (a, b: tuple[string, int]): int = cmp(b[1],a[1]))
03:21:09	reactormonk	Error: ':' or '=' expected, but found ']'
03:23:28	fowl	reactormonk, oh you have a tuple without types
03:23:54	fowl	tuple[key:string,val:int]
03:24:18	fowl	flaviu, how does an enum have less overhead than an int
03:24:20	reactormonk	figured
03:24:27	reactormonk	lib/system.nim(1882, 2) Error: undeclared identifier: 'result'
03:24:31	reactormonk	when using accumulateResult
03:24:32	flaviu	fowl: Enums are aliases for ints
03:25:18	flaviu	You can do type Foo = enum a=1, b=2, c=3
03:27:02	fowl	less overhead, where?
03:28:22	flaviu	I said 0 overhead compared to ints. They are basically alias for ints, so they have equal performance as an int.
03:28:40	fowl	eh
03:28:40	fowl	ok
03:29:33	flaviu	if an int's performance is 100%, an enum has 0 overhead, so 100%+(100*0) = 100%. Sorry if my wording was unclear
03:29:37	reactormonk	Error: cannot infer the type of the sequence
03:29:40	reactormonk	ah, sad story
03:29:42	*	tyler joined #nimrod
03:30:06	*	tyler is now known as Guest75193
03:30:31	*	Guest75193 is now known as TylerE
03:30:44	*	TylerE quit (Client Quit)
03:30:55	*	TylerE joined #nimrod
03:31:09	reactormonk	Error: expression 'sort(targets, proc (a, b: tuple[key: string, value: int]): int =
03:31:11	reactormonk	result = cmp(b[1], a[1]), Ascending)' has no type (or is ambiguous)
03:31:19	BitPuffin	fowl: you gonna test and merge or wat
03:32:54	fowl	reactormonk, sort sorts in place
03:36:59	*	flaviu quit (Quit: night)
03:37:21	fowl	BitPuffin, yes i keep getting distracted >_>
03:37:39	BitPuffin	fowl: hahah :)
03:41:32	reactormonk	I really wish I could just type .map(_.value) for getting something out of a tuple :-(
03:42:48	BitPuffin	do et
03:42:49	BitPuffin	naooo
03:44:34	fowl	reactormonk, probably possible
03:45:16	fowl	reactormonk, i think i can do that :D give me a few mins
03:46:21	BitPuffin	*ahem*
03:46:31	reactormonk	fowl, .mapIt(it.value) :-/
03:46:45	reactormonk	anyway, I would like targets.keys[0..1999].toSet to work
03:46:54	fowl	is that a thing? mapIT?
03:46:57	reactormonk	yup
03:47:09	fowl	i was going to write an &. macro
03:47:21	reactormonk	check sequtils
03:47:42	fowl	&.x would return proc(it:auto):auto=it.x
03:48:10	reactormonk	auto works?
03:48:16	fowl	mhm
03:48:22	fowl	BitPuffin, i already merge
03:48:23	fowl	d
03:49:35	*	askatasuna quit (Ping timeout: 258 seconds)
03:52:03	reactormonk	hmm, no del for counttables?
03:56:36	BitPuffin	fowl: did you test it?
03:58:34	*	brson quit (Quit: leaving)
03:58:56	reactormonk	apparently nope. hm...
04:09:25	fowl	BitPuffin, visually
04:14:06	Skrylar	wat
04:14:16	Skrylar	nimrod is trying to do boolean and instead of binary and
04:23:26	fowl	then give it numbers instead of bools lul
04:26:22	Skrylar	self.uint32 and button.uint32.ButtonMask
04:26:31	Skrylar	buttonmask(uint32): uint32
04:26:33	Skrylar	:|
04:27:59	fowl	did u import unsigned
04:31:12	Skrylar	yerp
04:33:38	Skrylar	time to go through the usual "something broke, update and rebuild compiler" dance :b
04:34:21	Skrylar	full compiler build in <20 seconds... i continue to be amazed
04:35:25	Skrylar	oh. well thats what went wrong
04:35:38	Skrylar	It doesn't understand that ButtonMask returns a uint32 and so it gets confused
04:38:59	reactormonk	muh, the sorting algorithm is tii slow
04:39:22	reactormonk	looks like I have to implement my own stuff
04:41:52	Skrylar	reactormonk: i did an insertion sort somewhere
04:42:05	Skrylar	what are you trying to sort?
04:43:41	reactormonk	Skrylar, a counttable, need top 2000
04:44:08	reactormonk	Skrylar, a modified insert sort should do
04:47:25	reactormonk	anything like mitems for seq?
04:48:59	reactormonk	nope.
05:02:06	Skrylar	i haven't benched my insertion sort, its a naive implementation thats like <30 lines
05:02:25	Skrylar	i wrote it because the rectangle packer for sprite sheets needs sorting, and its a small dataset
05:06:28	*	bjz quit (Ping timeout: 240 seconds)
05:09:19	reactormonk	hmmm
05:09:40	reactormonk	big dataset here
05:09:46	reactormonk	timsort, anyone? ^^
05:09:55	Skrylar	isn't quicksort the usual goto lol
05:10:09	reactormonk	sort() isn't quicksort
05:10:27	fowl	isnt it?
05:10:57	Skrylar	i didn't make any assertions on what sort() was, i said "isn't quicksort the usual goto" as in "isn't quicksort what people default to when insertion isn't good enough"
05:11:15	Skrylar	though i think in some fancy systems, the generic sort() function will check the total element number and then choose an appropriate subsort
05:12:14	reactormonk	fowl, nope
05:12:29	fowl	oh
05:12:36	reactormonk	Skrylar, timsort.
05:12:38	fowl	i thought we had a qsort()
05:12:43	reactormonk	fowl, where?
05:13:01	fowl	reactormonk, if its not in algorithms i was probably high
05:14:17	reactormonk	there's a few qsort things, but nothing accessible from nimrod
05:14:44	reactormonk	.gitignore:/tests/compile/tquicksort
05:14:46	reactormonk	hmmm
05:16:16	reactormonk	http://pastie.org/9083953
05:16:21	reactormonk	found this here in the git repo
05:16:48	fowl	reactormonk, oh apparently accumulateResult is supposed to be used in a function returning a seq
05:16:49	*	bjz joined #nimrod
05:17:15	fowl	reactormonk, i attempted to make it work more like a function but im not sure how to get the type of seq from the iterator
05:18:32	reactormonk	ohh, I use the sort of the counttables
05:18:35	reactormonk	which is shellsort...
05:20:28	fowl	reactormonk, if you still wanted to use accumulateresult in that way a better option is sequtils.toSeq
05:23:13	reactormonk	hmmm
05:23:17	reactormonk	does that work with iterators?
05:23:41	fowl	"Transforms any iterator into a sequence."
05:30:06	reactormonk	130648 <- number of elements in counttable
05:37:41	*	Demos quit (Read error: Connection reset by peer)
05:43:38	Skrylar	i don't even want to know why you have 130,000 objects in memory
05:43:53	Skrylar	but yeah, i don't think we have a sort for tables that big yet
05:55:30	Skrylar	reactormonk: i hope thats a table of indices or pointers
06:12:24	reactormonk	Skrylar, string -> int
06:12:33	reactormonk	Skrylar, some NLP models
06:12:51	reactormonk	looks like scala again...
06:22:01	Skrylar	is this a huge hash table?
06:23:07	*	nande joined #nimrod
06:30:49	*	noam joined #nimrod
06:33:16	*	noam_ quit (Ping timeout: 240 seconds)
06:39:49	reactormonk	Skrylar, yup
06:39:54	reactormonk	Skrylar, 100k, as mentioned.
06:44:32	Skrylar	you shouldnt need to sort a hashtable
06:45:42	reactormonk	Skrylar, I need the top 2k of the counttable
06:48:03	Skrylar	seems like there should be a neat way to use a rolling window or something and just iterate it
06:48:32	Skrylar	instead of trying to sort the whole 100k things, doing some cheaper check if there is room for something of a given size and keeping *that* sorted instead
06:49:38	Skrylar	with a lowest/highest counter so it can quickly skip over entries that have beneath the current minimum
06:50:41	reactormonk	Skrylar, wanna write it? ;-)
06:50:58	Skrylar	lol
06:55:45	reactormonk	Skrylar, hmm, priority queue would do it here
06:55:53	reactormonk	but I'm not sure if I want to implement a full one
06:59:50	reactormonk	how fast is sort of 2k elements?
07:05:46	Skrylar	well i acid test my hopscotch with 65k elements
07:06:13	Skrylar	so 2000 should be in the realm of a quicksort
07:06:34	Skrylar	that said you still have to iterate the whole word table
07:06:59	Skrylar	and then insertion sort the words in to the 2k area, removing the lowest ranked item if a newer one comes in
07:07:15	Skrylar	which is why i suggested storing the lowest, so you can skip most of the small cases
07:07:47	Araq	reactormonk: just use db_sqlite and call it a day
07:07:47	Skrylar	i might look at the data structure for that, but it won't be tonight
07:07:48	reactormonk	just the question where to insert the new one
07:07:58	reactormonk	Araq, hm?
07:08:07	Skrylar	reactormonk: in a skip list probably
07:08:33	Araq	I'll never understand why you keep all this data in fragile RAM, use a slim database instead
07:09:04	Araq	select top 2000 * from table order by someKey desc
07:09:24	reactormonk	Araq, using counttables mostly here
07:09:44	reactormonk	would you go for full SQL here?
07:09:48	Araq	what's the problem? shellsort not fast enough?
07:09:55	Skrylar	yay sqlite
07:10:13	Araq	"go for full SQL" is a misconception, db_sqlite is really sweet to use
07:10:34	reactormonk	Araq, yup
07:10:45	reactormonk	Araq, but I think go db_sqlite is a good idea
07:11:01	Araq	patch the sort to use the state of the art merge sort in algorithm.nim then
07:11:18	reactormonk	hmmm
07:11:30	Araq	though you need a "top 2000" which doesn't really require full sorting
07:11:51	reactormonk	indeed
07:11:59	reactormonk	but the idea of db_sqlite is good, I like it
07:12:25	Araq	you get serialization for free
07:12:38	reactormonk	yup
07:12:45	Araq	and if you don't need serialization it can't be serious work anyway ;-)
07:13:09	*	Araq <3 databases
07:13:24	*	Skrylar stores Araq in a database
07:13:26	reactormonk	Araq, :D
07:13:43	reactormonk	Araq, yup, I like to be able to serizalize
07:13:50	reactormonk	mostly because nimrod doesn't have a repl ^^
07:14:01	Skrylar	I guess it could have one with the VM
07:14:19	Araq	"nimrod i" does use the VM ...
07:14:37	reactormonk	Araq, yeah, any way to call that from yer code?
07:14:53	Araq	not yet
07:15:08	Araq	I lost the code :P
07:15:18	Araq	nah, not really but it's not ready
07:15:52	reactormonk	I wonder... should I store the corpus there too?
07:16:48	Araq	yes
07:17:08	reactormonk	but that's seq[seq[string]]
07:17:09	Araq	btw I will push the first implementation of spawn/sync tonight
07:17:26	Araq	so you can easily parallelize your code
07:17:44	reactormonk	hmmmm
07:17:59	reactormonk	Araq, got me some sample sqlite code around for init and table creation etc.
07:18:15	Araq	look at nimforum's code
07:18:22	Varriount	Araq: I'm nearly ready to open the PR for the getFileInfo code.
07:19:03	Araq	seq[seq[string]] usually requires some other data representation for SQL, reactormonk
07:20:09	Araq	the hacky way is to join(someSeparator) one level of 'seq' away ... :-)
07:20:50	reactormonk	Araq, pretty sure I don't want the corpus in here.
07:21:27	Araq	oh well
07:21:34	Varriount	How do I raise a generic error again?
07:23:01	Araq	raise newException(EWhatever, "some message")
07:26:19	*	DAddYE quit (Remote host closed the connection)
07:27:19	reactormonk	Araq, hmmmmm, now to figure out how to index stuff
07:28:05	Varriount	Araq: For tests, should I cause a failure by writing output, or by raising an exception?
07:29:47	Araq	Varriount: just let it crash somehow, make it output 'success' when successful and specify that as supposed 'output'
07:30:05	Araq	that's how I do it anyway
07:32:21	reactormonk	Araq, what's a data structure to store a few string -> string keys in? one or two, for column name -> value
07:33:39	Araq	create table Foo(key varchar(max), value varchar(max))
07:34:30	Araq	create index IX_Foo on Foo(key);
07:34:59	Araq	select value from Foo where key = 'lalala'
07:36:10	reactormonk	Araq, sorry. a nimrod data structure.
07:36:53	Araq	if you have very few keys use a seq[tuple[key, val: string]]
07:37:02	Araq	otherwise use a TTable[string, string]
07:37:33	Araq	and watch out that these are value based datatypes
07:37:52	Varriount	Araq: Submitted a PR
07:37:52	Araq	don't pass them around like no tomorrow as if it's a 'ref', it's not
07:39:18	Araq	Varriount: is it ready for inclusion into 0.9.4?
07:39:39	Varriount	Araq: Yes, to the best of my knowledge.
07:42:48	reactormonk	any way to escape for sqlite?
07:43:00	reactormonk	ah, good old ?
07:43:04	fowl	maaaan id like for tables to have an init() func so i didnt have to repeat the types of the table when i initialized it
07:43:56	Varriount	*I'd *didn't *I
07:44:41	fowl	ud like for me to use proper punctuals woldnt u
07:44:51	reactormonk	fowl, return type overload?
07:45:17	fowl	reactormonk, wat
07:45:27	Varriount	fowl: You're killing me.
07:46:55	reactormonk	Araq, aaaand how do I insert those tuples into sqlite queries? I don't really like plain insert, although I'm pretty sure the values don't need to be escaped
07:47:45	fowl	Varriount, i never knoticed before, you capitalize your sentences and use periods, lol, you arent writing a novel, this is IRC
07:48:10	reactormonk	fowl, this is freenode. A basic grammar is expected. Take your slang to quakenet.
07:48:30	fowl	reactormonk, *every* input into a db should be escaped even if you completely trust it
07:49:05	fowl	reactormonk, exec() does it for you
07:49:13	Varriount	Goodnight guys.
07:51:07	fowl	reactormonk, irt freenode, grammar, etc: i disagree
07:52:08	reactormonk	fowl, bit more complicated if you have varargs
07:52:50	Araq	reactormonk: db_sqlite does all the quoting for you
07:53:03	fowl	oh apparently all those functions take varargs
07:53:22	Araq	they all quote properly
07:53:27	reactormonk	Araq, now I have that seq[tuple[key, val: string]] - how do I escape that properly?
07:54:34	Araq	for tup in myseq: db.exec(sql"insert into Foo(key, value) values (?, ?)", tup.key, tup.val)
07:55:04	reactormonk	not that easy
07:55:29	reactormonk	each of the pairs determines which column has to match what
07:55:30	Araq	add some examples to the docs when you're done
07:56:16	reactormonk	looks like I have to
07:56:20	Araq	reactormonk: if it is not easy, you're doing it wrong
07:57:22	*	DAddYE joined #nimrod
07:57:40	reactormonk	Araq, trying to abstract over the update
07:57:56	reactormonk	ehh, inc, basically.
08:00:12	reactormonk	problem is, I got one or two columns as keys
08:00:34	reactormonk	Araq, http://pastie.org/9084295
08:01:40	*	DAddYE quit (Ping timeout: 240 seconds)
08:16:47	*	krusipo_ quit (Quit: leaving)
08:17:34	*	krusipo joined #nimrod
08:54:03	*	bjz quit (Ping timeout: 245 seconds)
08:57:55	*	DAddYE joined #nimrod
09:02:11	*	DAddYE quit (Ping timeout: 252 seconds)
09:09:12	*	bjz joined #nimrod
09:45:25	*	nande quit (Remote host closed the connection)
09:58:47	*	DAddYE joined #nimrod
10:03:21	*	DAddYE quit (Ping timeout: 250 seconds)
10:12:50	*	DAddYE joined #nimrod
10:17:31	*	DAddYE quit (Ping timeout: 258 seconds)
10:29:58	*	io2 joined #nimrod
11:13:35	*	DAddYE joined #nimrod
11:18:05	*	DAddYE quit (Ping timeout: 258 seconds)
11:24:53	NimBot	Araq/Nimrod devel c93c014 Dominik Picheta [+0 ±2 -0]: Modified tester output. Fixed JS tests.
12:14:31	*	DAddYE joined #nimrod
12:19:21	*	DAddYE quit (Ping timeout: 265 seconds)
12:40:43	*	DAddYE joined #nimrod
12:45:28	*	DAddYE quit (Ping timeout: 276 seconds)
12:47:25	BitPuffin	*sigh*
12:47:31	BitPuffin	koch install is useless
12:47:42	BitPuffin	at least on unixes
13:07:03	*	askatasuna joined #nimrod
13:32:55	*	askatasuna quit (Ping timeout: 276 seconds)
13:41:26	*	DAddYE joined #nimrod
13:43:07	BitPuffin	dom96, Araq: I have now created nimrod and babel packages for crux \o/
13:43:15	BitPuffin	just gotta publish them
13:46:34	*	DAddYE quit (Ping timeout: 276 seconds)
13:55:29	*	darkf quit (Quit: Leaving)
14:42:09	*	DAddYE joined #nimrod
14:43:03	*	Demos joined #nimrod
14:46:46	*	DAddYE quit (Ping timeout: 265 seconds)
14:54:49	*	BitPuffin quit (Ping timeout: 276 seconds)
15:02:05	*	shodan45 quit (Quit: Konversation terminated!)
15:42:47	*	DAddYE joined #nimrod
15:42:52	*	vendethiel quit (Read error: Connection reset by peer)
15:43:12	*	vendethiel joined #nimrod
15:47:21	*	DAddYE quit (Ping timeout: 252 seconds)
15:54:01	*	OrionPK joined #nimrod
15:54:26	*	Demos quit (Ping timeout: 265 seconds)
16:02:21	EXetoC	sup
16:04:41	*	[1]Endy joined #nimrod
16:04:43	Araq	I've implemented 'spawn'
16:06:36	EXetoC	ok how did you implement it?
16:06:54	Araq	as an AST->AST transformation in the compiler
16:06:57	*	Demos joined #nimrod
16:07:04	Araq	once the type API is ready we can make it a macro instead
16:07:45	EXetoC	great
16:07:50	Araq	still primitive though, doesn't do any work stealing
16:08:11	Araq	because work stealing makes no sense to me
16:08:55	Araq	I guess I am missing something :P
16:10:21	EXetoC	"Work-stealing has a serious advantage over work-requesting due to it's asynchronous nature: a thief thread is able to get some work, even if a victim thread is busy processing a user task or even de-scheduled by an OS..."
16:10:37	EXetoC	doing some googling now. seems relevant
16:10:48	EXetoC	"it's" misused again :<
16:16:35	Araq	well this doesn't mean anything though
16:18:22	Araq	it also completely ignores heap layouts and whether you have cheap sharing or not and message passing efficiency
16:22:19	*	joelmo joined #nimrod
16:22:26	*	joelmo left #nimrod (#nimrod)
16:24:22	*	[2]Endy joined #nimrod
16:28:12	*	[1]Endy quit (Ping timeout: 258 seconds)
16:32:30	*	[1]Endy joined #nimrod
16:35:40	*	[2]Endy quit (Ping timeout: 240 seconds)
16:43:32	*	DAddYE joined #nimrod
16:47:51	*	DAddYE quit (Ping timeout: 252 seconds)
16:51:51	*	[2]Endy joined #nimrod
16:52:44	*	Jesin quit (Quit: Leaving)
16:55:02	*	[1]Endy quit (Ping timeout: 258 seconds)
16:55:04	*	Demos quit (Ping timeout: 276 seconds)
17:02:13	*	[2]Endy quit (Quit: Want to be different? Try HydraIRC -> http://www.hydrairc.com <-)
17:10:05	*	BitPuffin joined #nimrod
17:12:53	*	askatasuna joined #nimrod
17:14:53	*	Demos joined #nimrod
17:20:20	*	Demos quit (Ping timeout: 258 seconds)
17:22:27	*	Demos joined #nimrod
17:28:49	*	Demos quit (Ping timeout: 250 seconds)
17:28:58	*	DAddYE joined #nimrod
17:28:59	*	q66 joined #nimrod
17:29:03	*	q66 quit (Changing host)
17:29:03	*	q66 joined #nimrod
17:40:28	*	filwit joined #nimrod
17:40:40	filwit	heya folks
17:41:01	*	q66 quit (Quit: Leaving)
17:41:06	filwit	i have some designs to show today :)
17:41:13	dom96	hellooo
17:41:15	dom96	yayyy
17:41:15	filwit	just finishing some graphics up first tho
17:41:25	filwit	give me an hour or so
17:41:53	filwit	ps. noticed the repo has been purged!
17:41:59	filwit	thanks reactormunk
17:42:01	filwit	monk*
17:42:52	dom96	filwit: We moved the release. New date is in 4 days.
17:43:00	filwit	oh, and I have a potential mascot too
17:43:01	*	q66 joined #nimrod
17:43:02	*	q66 quit (Changing host)
17:43:02	*	q66 joined #nimrod
17:43:07	filwit	dom96: still too soon
17:43:18	filwit	no way we can get everything done by then
17:43:30	filwit	at least, i doubt
17:44:13	filwit	maybe we can get something of the visuals in, but we should probably just plan on making another announcement latter
17:45:20	filwit	in fact it might better to separate them. I have new graphics for the website, forum, docs, and eventually the nimbuild too
17:46:03	filwit	so making a specific announcement for a "graphics design update" would probably stand on it's own
17:46:48	filwit	and the language updates kinda stand on their own
17:46:59	BitPuffin	woa fowl has the most epic name of all time
17:47:03	filwit	so we get more announcements this way
17:47:37	BitPuffin	I doubt we can fix all the stuff by then as well
17:47:38	*	Demos joined #nimrod
17:49:31	filwit	btw, do i use devel or purged_devel?
17:49:48	BitPuffin	probably devel
17:49:59	BitPuffin	I imagine purged_devel will become devel
17:50:19	filwit	well that did help, lol
17:50:26	filwit	which one is the actually purged one?
17:50:34	filwit	i imagine purged_devel
17:50:45	filwit	but i just saw devel get an update yesterday
17:50:47	filwit	so idk
17:51:01	filwit	...didnt** help...
17:51:23	BitPuffin	haha
17:51:25	BitPuffin	:P
17:52:14	BitPuffin	zahary: do you think the current issues with static[T] and generics can be ironed out in just 4 days? because it seems unlikely. And it doesn't seem wise to ship them in their current state
17:52:22	reactormonk	BitPuffin, nah, purged_devel can be killed
17:52:31	reactormonk	BitPuffin, it was just an intermediate experimet
17:52:42	BitPuffin	reactormonk: so in other words it has already become devel
17:52:44	filwit	thanks reactormonk
17:53:06	reactormonk	BitPuffin, just killed them to clear it up.
17:54:51	BitPuffin	reactormonk: you are so violent
17:55:50	BitPuffin	wow that is one big damn fast forward
17:57:00	dom96	brb
17:57:42	BitPuffin	We should set the release to like monday or something, and we'll spend the weekend in the VNUG helping each other fix compiler bugs
18:02:02	BitPuffin	filwit: would that be enough time?
18:03:44	Araq	hi filwit
18:03:52	filwit	hi Araq
18:04:00	BitPuffin	hey Araq
18:04:06	Araq	hi BitPuffin
18:04:07	filwit	i doubt it BitPuffin
18:04:13	BitPuffin	hrm
18:04:14	filwit	but maybe for just the art changes
18:04:19	filwit	i'll see
18:04:25	filwit	right now i have a lot of stuff to do
18:04:45	Araq	ping Varriount
18:04:53	BitPuffin	are you guys making money yet on the thing filwit?
18:06:01	BitPuffin	Araq: what do you think about a bug sprint?
18:06:12	filwit	BitPuffin, i told this to dom last time I was here, and I guess it's fine to say here. My Bones-Game project may have been shut-down, which is part of the reason why i haven't really been around.
18:06:32	BitPuffin	filwit: oh damn :(
18:06:33	filwit	BitPuffin: we got classified as a gambling app, and Google kick us off the Play Store
18:06:34	*	Demos quit (Ping timeout: 276 seconds)
18:06:43	BitPuffin	filwit: man that sucks
18:06:51	filwit	we're currently taking legal actions, but it's slow
18:06:53	BitPuffin	how are you guys recoverning?
18:06:56	BitPuffin	ah
18:07:00	filwit	and we just signed up a radio station too...
18:07:12	BitPuffin	:(
18:07:29	BitPuffin	filwit: hugs
18:07:33	*	gXen joined #nimrod
18:07:39	Araq	yeah my condolences
18:07:49	filwit	so things are a bit undetermined as far as my financial situation right now. But there's still hope too, so we'll see.
18:07:51	BitPuffin	hopefully it turns for the better
18:07:52	filwit	thanks, both
18:09:01	Araq	ok, guys here comes the important question
18:09:33	BitPuffin	guess this one should be labeled as feature https://github.com/Araq/Nimrod/issues/1095
18:10:15	Araq	we need an annotation/pragma for global variable usage tracking
18:10:15	filwit	after Araq asks his question, everyone please give me feedback about my mascot design: http://reign-studios.net/philipwitte/nimrod/mascot.png
18:10:38	Araq	rigth now I am torn between 'accesses' and 'uses'
18:10:54	filwit	vote: uses
18:11:06	Araq	it doesn't distinguish between read and write access
18:11:17	BitPuffin	I'd say uses as well
18:11:31	Araq	as that's not as useful as I thought and much more work to implement
18:11:37	Araq	ok, so
18:11:51	BitPuffin	filwit: looks like it could punch the hell out of some gophers :D
18:12:01	Araq	proc foo() {.uses: [stdout, stderr].}
18:12:02	filwit	lol
18:12:22	filwit	usesGlobs ?
18:12:29	filwit	usesGlob
18:12:34	BitPuffin	I feel like there is a better name
18:12:51	filwit	actually... 'glob' is kinda pointless i guess huh?
18:12:55	Araq	well it needs to be short :P
18:13:04	BitPuffin	uses is short
18:13:07	Araq	and 'glob' is not a common abbrev
18:13:10	filwit	yeah and you can't really 'use' a local anyways
18:13:19	BitPuffin	Araq: glob is a common abbrev
18:13:33	BitPuffin	although you'd confuse it with another kind of glob
18:13:34	Araq	BitPuffin: not in nimrod's stdlib
18:13:40	BitPuffin	Araq: exactly
18:14:16	BitPuffin	I think me and Araq have a more difficult time coming up with the word for it since we are not native englishers
18:14:17	filwit	Araq, is there, or can you make, a pragma which allows a proc to be sort of "thought as" a variable ?
18:14:36	BitPuffin	you mean kinda like js?
18:14:52	Araq	filwit: you mean invokable without () ?
18:14:55	filwit	Araq, when i want a proc accessor to a global property (so i down expose the original), i can't do Something.foo.bar
18:14:55	BitPuffin	where function foo() == var foo = function()
18:15:04	filwit	i have to do Something.foo().bar
18:15:07	filwit	Araq: yes
18:15:32	Araq	a template allows for that, filwit though this feature needs to be re-thought
18:15:57	Araq	my plan is
18:16:01	filwit	Araq: okay, well i'll come back to that later. It's just one of the two problems i've had recently with macro OOP stuff
18:16:16	Araq	template foo: expr = bar # no () for invokation necessary
18:16:25	Araq	template foo(): expr = bar # () for invokation necessary
18:16:32	BitPuffin	hmmmmmmmmmmmmmmmm
18:16:46	Araq	though this is against nimrod's philosophy
18:16:48	BitPuffin	will template foo: expr = bar still be callable with ()
18:16:50	BitPuffin	?
18:16:57	Araq	yeah, BitPuffin
18:17:00	BitPuffin	hm
18:17:04	BitPuffin	kinda inconsistent then
18:17:04	filwit	Araq: yeah it kinda goes against "write however, use however"
18:17:19	Araq	yup
18:17:19	filwit	Araq: i would almost prefer a pragma cause it's more explicit
18:17:21	BitPuffin	plus that makes it weird
18:17:25	BitPuffin	like in docs and stuff
18:17:39	BitPuffin	why can't they be used interchangably?
18:17:51	Araq	yeah well, I need to think about it
18:17:56	filwit	yeah
18:18:00	BitPuffin	yeah
18:18:08	filwit	btw, Araq, thoughts on my mascot proposal?
18:18:31	Araq	filwit: it's cool but too childish
18:18:33	BitPuffin	the gopherpuncher 2000
18:18:43	*	brson joined #nimrod
18:18:45	Araq	nobody will take us seriously with a mascot like that
18:19:02	BitPuffin	Araq: people take Go seriously
18:19:13	Araq	no they don't
18:19:14	BitPuffin	how about we make him hold a decapitated gopher head?
18:19:25	filwit	Araq: hmm.. okay. can you elaborate on what you'd like changed perhaps?
18:19:56	Araq	filwit: I'm not sure we need more than the crown we currently have, in fact
18:20:18	BitPuffin	that's where you are wrong
18:20:19	*	brson quit (Client Quit)
18:20:25	*	brson joined #nimrod
18:20:28	filwit	Araq: mascots are for TShirts and watermarks on the last page or presentations... the crown is the brand logo
18:20:30	BitPuffin	you need a nice cartoon animal to be successful these days
18:20:34	filwit	of**
18:20:56	Araq	well ok, IF we need an animal mascot
18:21:16	Araq	this is not it, it's not nearly friendly enough
18:21:19	BitPuffin	oh look, ocaml has a new website
18:21:28	BitPuffin	Araq: so what you're saying is more kawaii?
18:21:39	filwit	lol, Araq, you specifically told me to make him look angry last we talked about it
18:21:55	filwit	BitPuffin, lol @ kawaii
18:22:08	filwit	gunna draw an animu honeybadger now..
18:22:19	BitPuffin	do it
18:22:23	BitPuffin	with sparkly eyes
18:22:27	filwit	LOL
18:22:31	Araq	filwit: maybe but I consider it a strength to be able to change his mind
18:22:35	BitPuffin	and like, pink cheeks
18:22:44	filwit	Araq: fair enough
18:22:51	filwit	and i concur
18:22:56	BitPuffin	Araq: that's one fancy way of saying you can't make up your mind :P
18:23:27	filwit	but still, the mascot is one of the harder things to make, so it'll take a little while to change it, especially to make it completely different
18:23:30	BitPuffin	filwit: did you draw tihs in krita?
18:23:39	filwit	no, Inkscape
18:23:39	BitPuffin	or inkscape
18:23:43	BitPuffin	ah
18:23:45	BitPuffin	thought so
18:23:50	BitPuffin	since it looked kinda vectory
18:23:51	Araq	filwit: you only need to change its face though
18:24:18	Araq	maybe try to make it more like "honey badger don't care"
18:24:20	BitPuffin	how does it look if you put grey in the chest hair?
18:24:54	filwit	Araq: okay, "don't care" is descriptive enough. I'll see what i can do later
18:25:19	BitPuffin	I can think of a million ways to do don't care
18:25:20	Araq	but don't make it look like it's on drugs!
18:25:25	BitPuffin	you sure it's descriptive enough
18:25:36	Araq	and don't make it look like it's stupid
18:25:54	Araq	give him sunglasses, lol
18:26:06	BitPuffin	make him passed out with a bunch of beer bottles next to him and his wife and children look beaten up and are leaving him forever and have a text that says "honey badger doesn't care"
18:26:32	BitPuffin	or sunglasses.. yeah.. that works as well
18:26:46	BitPuffin	or WAIT!
18:27:08	BitPuffin	make a gopher that is holding on for dear life to a cliff, and have the honey badger walk away, WITH SUNGLASSES!
18:27:44	filwit	Araq: lol @ the drugs... will combine BitPuffins anime with a stoned badger now.. perfect mascot!
18:27:51	Araq	https://www.youtube.com/watch?v=718T3rkJR5o
18:28:15	filwit	also, sunglasses are not cool.. this is no longer the 90s
18:28:21	filwit	:P
18:28:36	Araq	I beg to differ
18:28:46	filwit	you still think it's the 90s?
18:29:01	Araq	I think they're still hip and trendy
18:29:02	filwit	:}
18:29:03	BitPuffin	sunglasses only make me think of http://youtu.be/X-LZWMc-OEY
18:29:11	BitPuffin	filwit: lol you troll :)
18:29:27	filwit	in florida we have to weir sunglasses to drive or the UV's hurt our eyes
18:30:23	filwit	okay, let me finish some stuff up and show y'all the rest of the designs
18:30:36	BitPuffin	80s hits are the complete opposite of cool Araq
18:30:44	Araq	well you better hurry with a better suggestion, BitPuffin
18:31:06	Araq	otherwise it's "uses"
18:31:10	BitPuffin	Araq: all of my suggestions were better
18:31:15	BitPuffin	oh
18:31:17	BitPuffin	you mean to uses
18:31:21	BitPuffin	well I can't think of one
18:31:26	BitPuffin	but it feels like there's something better
18:31:44	BitPuffin	however I imagine uses is the shortest
18:31:59	BitPuffin	when will this be needed btw?
18:32:25	Araq	BitPuffin: we have a new notion of "thread safety"
18:32:52	filwit	new notion: "safety is for looser squares"
18:33:10	Araq	where "thread safe" == "uses only globals that do not contain GC'ed data"
18:34:39	dom96	back
18:35:32	dom96	filwit: Really cool mascot. But I agree with others, it shouldn't look pissed off.
18:35:51	filwit	k, thanks for feedback
18:36:23	dom96	filwit: But do save it. Maybe someday we will use it for something :)
18:36:51	dom96	Araq: So 'uses' means "this global is used in a thread"?
18:39:39	Araq	no it simpy means proc uses global
18:43:45	dom96	why not call it 'usesGlobal' then?
18:44:43	Araq	doesn't feel right
18:46:04	filwit	dom96, i suggested the same thing, but can you even 'use' a local really?
18:46:46	filwit	and i thought you already had read/writes ?
18:47:54	Araq	no but it is planned
18:48:33	filwit	well i know it's not done, i guess i'm asking what the difference is between 'read/write' and 'uses'
18:50:12	Araq	well 'uses: X' means 'reads: X; writes: X' obviously
18:51:01	filwit	k
18:52:15	dom96	how about 'in'?
18:52:44	Araq	for "input"?
18:52:51	dom96	yeah
18:53:11	BitPuffin	Araq: how is using globals that don't contain GC'd data thread *safe*
18:53:12	filwit	'uses' makes more sense for what it does i think
18:53:18	BitPuffin	yeah
18:53:23	BitPuffin	I hate in and out keywords
18:53:35	dom96	'in' is already a keyword lol
18:53:44	BitPuffin	it is?
18:53:47	BitPuffin	I've never used it
18:53:52	filwit	:|
18:53:58	filwit	it's used everywhere in Nimrod..
18:54:08	BitPuffin	wat
18:54:12	filwit	if Something in { A, B, C }:
18:54:15	BitPuffin	oh
18:54:18	BitPuffin	well that is fine
18:54:24	BitPuffin	I meant more like for function parameters
18:54:26	BitPuffin	like in D
18:54:27	filwit	for x in list: ...
18:54:28	BitPuffin	they had like
18:54:30	BitPuffin	in
18:54:32	BitPuffin	out
18:54:33	filwit	yeah
18:54:34	BitPuffin	inout
18:54:36	BitPuffin	never really got what it was for
18:54:46	BitPuffin	for x in arr is totally cool
18:54:47	filwit	inout was for const stuff
18:55:10	filwit	in was basically the same as 'const', and out is the same as C#'s out
18:55:51	BitPuffin	ah
18:55:55	BitPuffin	well then why not just const
18:55:57	BitPuffin	inout
18:56:08	filwit	inout means you can pass either a const or not
18:56:10	BitPuffin	the fuck kind of a way is that to describe a constant
18:56:24	BitPuffin	filwit: well in C you can also pass a const or not
18:56:27	filwit	i'm really not sure of the difference between 'in int' and 'const int' but i think there is one
18:56:33	BitPuffin	yeah
18:56:37	BitPuffin	that's why I was confused
18:56:39	BitPuffin	it's stopid
18:56:39	filwit	'out int' means the variable is ref, but created in the proc
18:56:41	BitPuffin	anyway
18:56:47	filwit	yeah
18:56:57	filwit	can be confusing
18:57:06	filwit	const in C is worse in some ways though
18:57:08	BitPuffin	Araq: should you really be thinking of this now though? Probably better to focus on fixing bugs for release
18:57:12	filwit	cause placement is important
18:58:47	Araq	BitPuffin: upcoming release has too few killer features
18:59:04	Araq	so I'm adding the major parts of the new concurrency model
18:59:23	BitPuffin	Araq: in 4 days? ._.
18:59:27	Araq	yes.
18:59:48	Araq	I've already added 'shared ptr T' and 'spawn' and 'sync'
19:00:24	Araq	global tracking is trivial to implement
19:00:39	BitPuffin	Araq: yes but it needs extensive testing
19:00:56	Araq	for 0.9.6 only 'guarded' and deadlock freedom are left then
19:01:44	dom96	Araq: What about async? :(
19:02:25	Araq	dom96: I'm still trying to fix it for the release
19:03:07	Araq	but it could be a time grave and then we have no killer feature for 0.9.4
19:03:27	dom96	Async await isn't a killer feature?
19:03:44	Araq	oh it is.
19:04:11	Araq	but it's too risky to only focus on that
19:04:47	Araq	so I'm doing both things instead
19:05:03	renesac	another pragma name suggestion: 'rwGlobals'
19:05:10	renesac	but I'm not sure if it is better than 'uses'
19:05:29	BitPuffin	dom96: maybe YOU should focus on it ;)
19:05:32	BitPuffin	seriously though guys
19:05:36	BitPuffin	bugsprint over the weekend?
19:05:56	renesac	Araq, are you sure distinction between reads and writes isn't needed?
19:06:13	renesac	a shared global state only for reading data is much safer than for rw data
19:07:19	dom96	BitPuffin: I don't have the knowledge or the time.
19:07:26	Araq	renesac: when you have thread local heaps reading is actually not as harmless as it looks
19:07:32	BitPuffin	dom96: acquire
19:07:45	Skrylar	meep
19:08:04	BitPuffin	Araq: how can you implement uses when reads and writes hasn't been implemented
19:08:16	BitPuffin	if uses = reads + writes
19:08:39	Skrylar	it may just not make a distinction between the two
19:08:57	Skrylar	"this is a pointer therefore it is used" vs "i have analyzed the pointer is only read"
19:09:02	BitPuffin	dom96: what about if you look at it during a bug sprint tho
19:10:11	*	nande joined #nimrod
19:10:57	Araq	BitPuffin: what Skrylar said is correct
19:11:10	renesac	Araq, sure many things can still go wrong, but if the initial setup and latter destruction is well separated phase from the threaded code, not so much
19:12:08	BitPuffin	Skrylar, Araq: Ah, so it's basically saying that it both reads and writes without actually verifying it?
19:12:25	renesac	and I laughed a bit in nimrod's new notion of thread safety, but I understand where this come from
19:12:26	renesac	XD
19:12:26	Araq	BitPuffin: yes, it only says it 'uses' it
19:12:46	BitPuffin	then I get why it is more tirvial
19:12:50	Skrylar	renesac: new notion of thread safety?
19:12:59	Skrylar	did we get the concurrency model of a clown hat
19:13:56	BitPuffin	still have a hard time of seeing why this gives thread safety
19:14:02	renesac	Skrylar, [15:32:25] <Araq> BitPuffin: we have a new notion of "thread safety" [15:33:10] <Araq> where "thread safe" == "uses only globals that do not contain GC'ed data"
19:14:20	Skrylar	i thought it already did that
19:14:32	BitPuffin	it does
19:14:43	Araq	renesac: for a purely RC'ed GC any read operation is an RC update and thus racy unless it performs an atomic operation
19:14:43	BitPuffin	but it's not verified I guess
19:15:00	renesac	Araq, yeah
19:15:06	Skrylar	fortunately atomic operations for ints are in most of the compilers
19:15:17	BitPuffin	Araq: I believe the correct term is "racist"
19:15:37	renesac	thread safe GCs seems a very complex beast, unless you go the thread local heaps way like most new languages do, including nimrod
19:15:46	*	BitPuffin hides in a corner
19:15:47	Skrylar	operating only on static globals doesn't guarantee thread-safety, actually
19:16:01	Araq	yes, yes, I know
19:16:04	Skrylar	it prevents the RC racing that Araq mentioned, it doesn't do anything to stop two threads independently clobbering a global
19:16:23	renesac	yep, that is why it was funny
19:16:25	Araq	yes that's why will get more checking for globals later
19:16:27	Skrylar	renesac: threads are pain :(
19:16:39	BitPuffin	yeah
19:16:41	BitPuffin	well
19:16:46	BitPuffin	I mean
19:16:54	Skrylar	microthreading is <3, multithreading is </3
19:17:11	BitPuffin	if we get as strong thread local stuff as other languages and a nice global system like this I think we're pretty good
19:17:36	BitPuffin	Skrylar: isn't microthreading fully transparenty though?
19:17:37	Araq	this rule patches a major hole that we have right now, it doesn't fix everything
19:17:50	Araq	read my blog for more details
19:18:00	BitPuffin	Araq: so right now there is nothing stopping me from using a ref as a global?
19:18:12	renesac	but does microthreading works for anything not trivially paralelizable/embarasingly parallel?
19:18:35	Skrylar	renesac: no, microthreading is used to maximize use of a single core
19:18:46	BitPuffin	aka hyperthreading?
19:18:54	Skrylar	no, hyperthreading pretends to be an extra cpu core
19:18:56	renesac	hum, to mask cache misses?
19:18:57	BitPuffin	or do you just mean the concurrency the cpu does automatically
19:19:02	Araq	there is a shitty global analysis pass that tries to ensure safety, BitPuffin
19:19:13	BitPuffin	Araq: ah, but it's easy to get around
19:19:14	Araq	but I will remove it
19:19:34	Skrylar	microthreading is useful in some cases like dealing with I/O and futures
19:19:49	Skrylar	since you can go "hmm there aren't any more packets yet? jmp to somewhere else instead of waiting"
19:20:08	Araq	Skrylar: we have that with async/await
19:20:20	Skrylar	isn't async/await unfinished :P
19:20:31	Araq	it works but leaks memory :P
19:20:39	Skrylar	then it doesn't work :F
19:20:50	Araq	ha, you have no idea
19:21:24	Araq	rails used to leak like no tomorrow, people used it in production nevertheless
19:21:39	Araq	ok, the Rails guys are special, but still
19:22:05	Skrylar	since when do rubyists care about things, its one of the slowest interpreters in the world xD
19:22:22	Araq	not anymore
19:22:24	renesac	Skrylar, it got faster
19:22:40	Skrylar	i haven't used ruby 2x
19:22:58	Skrylar	i remember they were trying to build rubinius and a few others to make the VM uncrappy
19:23:11	Skrylar	I bet luajit still wins :)
19:23:42	Araq	I bet it doesn't for GC heavy benchmarks
19:25:24	*	Varriount|Mobile joined #nimrod
19:26:21	*	nande quit (Read error: Connection reset by peer)
19:26:42	*	nande joined #nimrod
19:31:21	reactormonk	Araq, I'm special \o/
19:43:48	BitPuffin	look at Araq acknowledging good things about ruby
19:43:58	BitPuffin	spelling.se
19:44:38	dom96	I wonder if Yukihiro Matsumoto has a gun pointed to his head.
19:46:25	Skrylar	dom96: why would he? other people were writing VMs
19:46:52	Skrylar	matz always said ruby was about being fun to program in above all else
19:48:49	Varriount|Mobile	Araq: You rang?
19:49:51	Araq	Varriount|Mobile: how come even mingw gcc 4.8 doesn't support __sync_bool_compare_and_swap ?
19:52:31	Varriount|Mobile	Araq: No idea
19:52:50	Araq	it's really annoying to have no atomic ops for win32
19:53:55	Varriount|Mobile	Araq: Have you tried asking #mingw?
19:57:24	Varriount|Mobile	http://sourceforge.net/p/mingw/bugs/1585/
19:57:59	fowl	BitPuffin, i have another allegro test for u to run later (if u dont mind)
19:59:57	BitPuffin	fowl: HOW DARE YOU
20:00:39	BitPuffin	No but who could ever say no to someone called Billingsly Wetherfordshire
20:00:46	BitPuffin	too cool name to conceive
20:01:09	*	Matthias247 joined #nimrod
20:05:16	dom96	oh cool. `%` supports named vars
20:05:27	dom96	There should really be an overload which accepts a string table.
20:06:13	Araq	I think there is
20:06:21	Araq	it's in strtabs though
20:07:00	dom96	oh cool
20:08:20	Varriount|Mobile	Araq: I think sync_bool_compare_and_swap is available on mingw-w64
20:13:39	*	Mat3 joined #nimrod
20:13:46	Mat3	hi all
20:13:56	Araq	hi Mat3
20:14:11	Mat3	hi Araq, what's new ?
20:14:33	Varriount|Mobile	filwit: I love the mascot! It really displays nimrod's "I'm a strong language and I don't care" attitude.
20:14:51	filwit	thanks for the feedback, Varriount|Mobile
20:15:29	Araq	Mat3: I've implemented 'spawn'
20:15:45	Mat3	great !
20:16:36	Araq	speaking of which, how do you measure low level cycle counting stuff?
20:17:20	Mat3	I use the performance counters which depend on the processor type
20:17:41	Araq	ok
20:21:34	Mat3	you will likely need three measures, one for the number of cycles, one for possible BTB misses and at last one for the number of cache misses
20:26:36	NimBot	Araq/Nimrod devel 6692d95 Dominik Picheta [+0 ±1 -0]: Fixes incorrect nodejs detection in tester.
20:26:36	NimBot	Araq/Nimrod devel 2fb5d62 Dominik Picheta [+0 ±24 -0]: Tester now appreciates the test target. Modified 'cmd' in specs.
20:27:06	*	wan joined #nimrod
20:27:22	Araq	hi wan
20:27:31	wan	hi
20:29:07	*	BitPuffin resist saying joke about being the chosen wan
20:29:12	wan	concerning the mascot, I think it looks too fat. It's a bear ! (or a fat dog...). Although honey badger have a loose skin, their arms are definitely slimmer
20:29:39	wan	*legs too
20:30:37	fowl	BitPuffin, :D
20:31:06	wan	The gradient on the crown is a little out of place, there is no other gradient in the character
20:31:32	BitPuffin	wan: the charecter and crown is not the same material either though so that doesn't make sense to say
20:32:34	fowl	dom96, i looked into making it allow dictionaries but the implementation is pretty involved
20:33:01	fowl	dom96, i mean "$x" % {"x":"hello"}
20:33:17	wan	BitPuffin: It does, however. It's like mixin flat design with web2.0 gradients, it doesn't mix that well
20:33:53	wan	especially considering the 'comics' look of shadows for the honey badger
20:34:22	wan	(fat pencil in the shadows zones)
20:34:30	dom96	fowl: strtabs does indeed have its own %
20:35:48	fowl	yea but i dont generally need to start a strtab for the input
20:35:55	Mat3	hmm, what mascot ?
20:36:14	dom96	fowl: ahh indeed.
20:36:30	dom96	fowl: So what was the problem, too much work?
20:36:44	fowl	".." % ["x","hello"] works but {"x":"hello"} is actually [(key:"x", val:"hello")] so the implementation for it would be simpler than the current way
20:37:17	Araq	yup, fowl is right
20:37:29	fowl	dom96, well the current way (looks like) allows you mix $1 $2 $x
20:38:06	Araq	fowl: I don't think anybody used that feature though
20:38:07	fowl	thinking about it again, it doesnt sound difficult anymore, it should be a separate function than the other %
20:38:33	EXetoC	Araq: the dictionary sugar?
20:38:35	Araq	Mat3: please read the logs, filwit created a new mascot
20:38:43	EXetoC	%?
20:38:53	Araq	EXetoC: no, mixing $1 and $name
20:38:56	filwit	Mat3: http://reign-studios.net/philipwitte/nimrod/mascot.png
20:39:26	fowl	filwit, i like the color scheme on the far right
20:39:30	fowl	reminds me of gobo :>
20:40:00	filwit	yeah i like that one two.. fits into the a slide on the homepage well
20:40:05	filwit	:)
20:40:10	filwit	just about ready to show
20:40:23	filwit	kinda rushed some last graphics, but you folks'll get the idea
20:41:11	fowl	http://distrowatch.com/images/yvzhuwbpy/gobo.png
20:41:32	Mat3	filwit: I can not access the server
20:41:48	filwit	Mat3 really, odd...
20:41:58	filwit	it's my own server, try again in a minute
20:42:01	wan	filwit: I realized I didn't say anything positive about your mascot. I shall have! I do like the all-grey look. I'm only good at criticizing and noticing defaults, though, don't take it the wrong way
20:42:25	filwit	wan, i missed all comments from you before, sorry :S
20:42:42	filwit	wan, wasn't paying attention, but i'll rescan logs and look in a minute
20:43:08	filwit	wan, also, thanks for the feedback, positive or not
20:45:13	filwit	wan, ah, i see "too fat". Yeah i was trying to slim him down a bit, and i can do that more. I was using these as reference: http://ngm.nationalgeographic.com/ngm/0409/feature6/images/ft_hdr.6.jpg http://www.badassoftheweek.com/honeybadger1.jpg
20:45:37	wan	Has anyone come accross a 'web templates' library that's nimrod-compatible? Like Mustache, something that exists for other languages as well (not like source code filters)?
20:48:05	wan	'cause I'll soon have something for jade templates (compiling them to nimrod code), don't know if anyone would be interested
20:48:56	dom96	wan: OrionPK created a templates thing
20:50:14	wan	Is it on github somewhere? I guess the username is not the same. Do you have it?
20:50:28	dom96	I think this is it: https://github.com/onionhammer/onion-nimrod/tree/master/templates
20:51:32	wan	Thanks, I'll see how he does the calling. Could be interesting to have a common convention
20:54:57	OrionPK	it doesnt work on devel for some reason though
20:55:06	OrionPK	idk if it's a regression of what
20:55:08	OrionPK	or*
20:55:22	OrionPK	my templates work well with NimLime wan
20:56:10	Araq	OrionPK: thanks for reporting that
20:56:26	OrionPK	https://dl.dropboxusercontent.com/u/417554/logic_example.png
20:56:31	OrionPK	araq lol sorry
20:56:39	OrionPK	araq I've been super busy last...... 4 weeks
20:56:43	wan	Ah, that solves the syntax highlighting for strings inside nimrod files, nice. I'm on vim though, so not for me, but that feature is cool.
20:56:57	wan	Didn't know about NimLime
20:57:30	OrionPK	right, the html""" """ prefix just tells sublimetext what it is
20:57:32	OrionPK	how to highlight it
20:57:56	OrionPK	the tmpl/i bit translates it into result.add's and nimrod logic at compile time
20:59:27	wan	For interfacing, I was planning on doing something like `proc my_template(locals): string = ...` (haven't tested yet)
21:00:02	wan	Because I want the templates in separate files (.jade), I don't want to declare what they take as parameters
21:02:09	OrionPK	why have the templates in separate files?
21:02:25	OrionPK	I guess it's nice if you want to hcange them w/o having to recompile your nimrod
21:02:30	OrionPK	but then you have runtime costs associated
21:02:32	Mat3	filwit: how about a sword as logo (because Nimrod is secure) ?
21:02:32	wan	syntax highlighting was one reason, reusing them for client-side is another
21:02:50	EXetoC	OrionPK: you can cache it
21:03:27	filwit	Mat3: pretty sure logo wont change to something completely different :) and mascots are usually a creature of some kind, not an object
21:03:42	EXetoC	in some cases
21:03:58	wan	runtime costs? No, my_tempalte.jade->my_template.nim in a makefile, then `import my_template` in nimrod
21:04:51	Mat3	filwit: how about a frog with a crown ?
21:05:14	Mat3	(because Nimrod share so much hidden features)
21:05:16	wan	OrionPK: so I still have to recompile when I change my templates
21:05:16	EXetoC	a snail
21:07:12	Mat3	the frog is also the enblem of the merowinger dynasty
21:07:14	filwit	Mat3: the honeybadger wasn't chosen by me, but by someone else, and liked by Araq
21:08:19	OrionPK	wan ah then probably my templates are better :P
21:09:52	wan	OrionPK: well, I'll never miss closing a tag with mine B)
21:10:15	EXetoC	can doc comments be positioned below constants and fields?
21:12:07	Araq	EXetoC: use #\
21:12:13	Araq	# it like this
21:12:13	OrionPK	heh if you forget in mine, it wont compile
21:13:44	wan	OrionPK: I really want to try the philosophy of sharing templates between client and server, not having to either recode your templates or make javascript necessary. That was my goal for porting (most of) jade
21:14:16	OrionPK	it's an interesting idea
21:15:09	wan	The problem being that you need to use the same template engine on both sides, and the choice is limited. Highly dependent on your stack...
21:17:06	Araq	how do you share the template without javascript on the client?
21:17:48	EXetoC	clients? templates?
21:17:57	OrionPK	use html on the server araq
21:17:59	OrionPK	:p
21:18:05	wan	You serve already-filled html to the client, the ones with javascript can then dl scripts and pre-compiled templates, and query json api for an enhanced experience
21:18:57	wan	I'm designing my server-side views so that each one answers to json and html
21:19:34	wan	perfect marriage of webapp and website
21:19:47	Araq	hmm I know the words you're using but I don't get anything
21:19:49	wan	no compromise on SEO
21:20:27	wan	SEO* I mean crawlability
21:20:29	OrionPK	yeah im not sure I see where you're going w/ it wan
21:21:08	OrionPK	why not just have the clients download the precompiled HTML and serve it rather than use JSON
21:21:26	*	gXen quit ()
21:22:25	wan	JSON saves on bandwith when you have already fetched the pre-compiled templates.
21:22:52	OrionPK	JSON means more client-side processing
21:23:05	OrionPK	to translate what came back into HTML
21:23:08	OrionPK	and then render that
21:23:29	wan	I'm betting that phones have enough power to do that well, 3G connection is still more of a problem for latency nowadays
21:23:31	fowl	better than overheating the server
21:23:35	OrionPK	lol
21:23:56	OrionPK	fowl I saw you (or someone) fixed that dependency in the sdl lib, thanks
21:23:59	wan	Yeah, the server doesn't care, it's not that more expensive to serve html
21:24:15	fowl	OrionPK, gracias pointing it out
21:24:35	OrionPK	i got the skeleton running on my mac :D
21:24:53	OrionPK	(again)
21:25:58	Araq	wan: serving static html can go through a highly optimized path in nginx for instance
21:26:20	Araq	so the server does care
21:26:35	*	Clement joined #nimrod
21:26:43	Araq	hi Clement welcome
21:26:58	Clement	Hello Araq
21:26:59	*	Clement is now known as Guest87695
21:27:33	wan	Oh yes, but this is for static files. Templates are for when your pages are dynamic, of course.
21:28:26	Guest87695	Could you please tell me whether this is a bug of an intended behavior ? https://gist.github.com/ClementJnc/a31a8dcf61668e2d045e
21:28:27	EXetoC	not only
21:30:58	dom96	Guest87695: Works for me. Are you using 0.9.2?
21:31:11	Guest87695	I have different behavior with "result &= ..." and with "result = result & ..."
21:31:26	Guest87695	Using the last version from Git.
21:31:50	fowl	how about result.add ".."
21:32:00	dom96	oh, I see what you mean.
21:32:04	dom96	Thought it crashed.
21:33:05	Guest87695	result.add give the same behavior as &=
21:34:53	Araq	Guest87695: is that a regression you hunted down to a minimal snippet?
21:35:18	Guest87695	Yes
21:35:51	wan	regression means it worked before, then?
21:36:15	Araq	report it please. Oh and we need a 'regression' tag.
21:36:48	Araq	wan: that's what 'regression' means, yeah. And these are our biggest problem.
21:36:59	Guest87695	OK, I'll prepare a report.
21:38:32	wan	I didn't know the compiler changed that much. I guess after a while, you'll have a nice collection of tests if things break more often.
21:38:41	*	Demos joined #nimrod
21:39:07	wan	Dirty trick for getting test contributions: break things.
21:39:22	wan	I'm joking of course
21:39:40	Guest87695	In fact, I have to check, I'm not sure this exact version has worked with an older version.
21:42:26	wan	(That's why I asked about regression (I know what it means!). Araq's question had two components: is it a regression, and have you made it minimal. I sensed that the singular response might need some double checking)
21:43:28	Demos	wan: protip -- you can use git bisect to quickly find out if it is a regression
21:43:35	Demos	well fairly quickly
21:43:46	Demos	it helps that nimrod compiles really fast
21:43:54	Araq	wan: between 0.9.2 and the upcoming 0.9.4 we have over 1300 changes
21:44:18	Araq	so much for the "compiler doesn't change that much"
21:44:29	Araq	:D
21:44:37	EXetoC	c(:)-<
21:44:43	wan	Demos: does it recompiles backward for each commit and stops when it works? That seems awesome, didn't know about bisect!
21:45:32	Demos	it is a git feature, you give it a known working commit and a known buggy commit and it will checkout the commit between them, you test, then you tell git if it worked or not
21:45:49	Demos	you can also give git a script that does the test and reports the result (by exit code I think)
21:46:16	Demos	it really pushes home how cool log(n) algorithms are
21:47:24	wan	binary search ftw?
21:47:35	Demos	pretty much :D
21:48:24	Demos	a fun project would be to integrate that into the builder, so you could add a test and then see when it started failing in the past
21:48:51	wan	Doesn't the history purge kinda limit that?
21:49:00	Demos	no
21:49:06	*	faassen left #nimrod (#nimrod)
21:49:21	Demos	the history is still there, all the hashes just changes since we deleted some large binaries and c sources
21:49:28	Demos	*changed
21:49:52	filwit	Okay, due to an oversight on my part, I wont be showing the Web/Forum/Docs designs publicly. However, if you'd like to see them, please PM me and I'll see about showing them to you.
21:51:25	wan	Nice futuristic look, however MASSIVE lack of contrast
21:51:42	filwit	the designs are ready to show, and i'd appriciate any feedback, so please let me know if you're interested
21:51:44	wan	you need the main background to be whiter
21:51:58	wan	(already found the links without asking you)
21:53:22	wan	you might want to turn up the opacity on that #fff rgba that I suspect you have there
21:54:00	wan	the grey-on-blue seems nice though
21:54:10	dom96	I thought the same thing. I wonder if filwit's monitor's brightness is set to "THE SUN".
21:54:44	filwit	wan... damn you!
21:54:51	filwit	but yes, that's the general idea
21:54:56	*	Guest87695 quit (Quit: Page closed)
21:55:05	filwit	also, not everything is fully fleshed out obviously
21:55:13	BitPuffin	YOU KNOW WHAT
21:55:14	Mat3	need some sleep, ciao
21:55:19	*	Mat3 quit (Quit: Verlassend)
21:55:19	BitPuffin	Desperate times
21:55:22	BitPuffin	desperate measures
21:55:23	wan	filwit: Oh, did you mean you prefer feedback to be given in pm?
21:56:27	filwit	wan, it's okay, just that we don't want to spread the idea of a logo change, and don't want random people to think it's changing due to these images
21:56:29	BitPuffin	for some reason doing this feels like admitting failure
21:56:31	BitPuffin	but it's not really
21:57:17	wan	BitPuffin: care to share your worries?
21:58:12	BitPuffin	wan: well since bugs with generics is causing my 3d math lib not to work at all I am developing a "boring" subset of the library
21:58:31	BitPuffin	ie one that is only generic with the type of the vector/matrix components and not size
21:59:15	Demos	I just converted some of my functions to have names based on the size and type, this way I can convert back eaisly when the compiler supports static[T] better
21:59:18	BitPuffin	so the boring subset will basically be just like every 3d math lib for every language whereas the regular ones that don't work are generally moar powerful than what is offered
21:59:39	EXetoC	it's temporary so whatever
22:00:17	wan	filwit: on the docs page, the docstring below procs is too grey, need to be blacker
22:00:46	reactormonk	proc createModel(db: TDBConn, cc: ContextCounters, into: string) =
22:00:47	wan	filwit: otherwise that one looks fantastic
22:00:53	reactormonk	expression 'createModel(db, cc, "basic")' has no type (or is ambiguous)
22:00:58	filwit	wan, yes the docs are generally not contrasted enough
22:01:04	reactormonk	oh, it doesn't have a return type
22:01:11	filwit	wan: this will be fixed once i turn it into HTML
22:02:07	wan	filwit: I was talking about docs.htm, not the pictures this time
22:02:14	filwit	oh i see
22:02:25	filwit	docs.htm is not a graphics design thing
22:02:40	filwit	colors where chosen to be consistent, not production
22:03:12	filwit	it's just to illustrate different things, like when you click on a type, it isolates all it's procs... the sidebar is togglble, etc...
22:04:30	wan	Oh yeah. Much wow.
22:05:02	dom96	BitPuffin: Just implement what you need instead of a whole damn library.
22:05:33	wan	You can improve the colors if you want, the design is such a great step forward compared to the current docs...
22:05:57	filwit	great!
22:06:05	filwit	thanks for your input, wan
22:06:10	dom96	yeah, in fact I would say it's better than any docs i've ever seen.
22:06:33	wan	filwit: happy to help
22:08:03	BitPuffin	dom96: well I want it to be available for others as well
22:09:08	fowl	thats what we have SVN for
22:09:39	BitPuffin	fowl: you are drunk?
22:12:51	wan	filwit: (back to the pictures)the forum is a bit bland. Maybe expand each line's height? You could try having the posts and/or views counts on the left, a bit separated (in their own box) a la stackoverflow, or even a la HackerNews.
22:13:03	BitPuffin	dom96: we should really *really* fix the security derps with the forum before the new release btw
22:13:15	BitPuffin	if this brings a wave of new users it's simply unacceptable to keep doing this
22:13:22	dom96	BitPuffin: are you volunteering?
22:13:40	BitPuffin	maybe yeah
22:13:51	BitPuffin	just that jester doesn't really work on mac
22:13:54	*	Matthias247 quit (Read error: Connection reset by peer)
22:14:02	filwit	wan, thanks. I'll experiment more once this is in HTML/CSS and we can play with those sorts of things
22:14:06	dom96	BitPuffin: why not?
22:14:13	BitPuffin	dom96: redirect
22:14:29	dom96	BitPuffin: that's a compiler bug right?
22:14:35	BitPuffin	aye
22:14:43	fowl	BitPuffin, yep
22:14:47	wan	filwit: we need to break up this 'lines, lines everywhere' aspect of the forum. Maybe introduce images on the left, a la forum.nofrag.com/forums/nofrag ?
22:14:53	dom96	gosh, so many problems with them Macs.
22:14:59	BitPuffin	fowl: cool, what are you drinking?
22:15:02	BitPuffin	dom96: agreed
22:15:09	*	brihat joined #nimrod
22:15:28	fowl	american beer!
22:15:35	dom96	BitPuffin: Could you test async await on mac?
22:15:56	BitPuffin	dom96: signed cookies would be acceptable compared to sending the password hash back and forth, but I'd really prefer the session store since that's the real solution
22:16:09	BitPuffin	however changing the hashing is probably prio #1
22:16:16	filwit	wan, pointless pics don't add much though... maybe show a small icon of the OP's avatar or something tho
22:16:25	BitPuffin	dom96: the easiest way to migrate would probably be to have everyone reset their password
22:16:54	EXetoC	password recovery next then
22:17:02	BitPuffin	EXetoC: not necessarily
22:17:29	BitPuffin	EXetoC: could be first a login attempt with the new hash, if that fails, try with md5 and compare, if successful, ask them to write new password
22:17:59	BitPuffin	dom96: I could, but you'd have to show me the way
22:18:57	*	flaviu joined #nimrod
22:19:15	fowl	did u all put heartbleed bugs in ur softwares yet?
22:19:17	BitPuffin	fowl: local brew?
22:19:23	wan	filwit: yeah, I know, useless pics are useless. Well, if you feel like adapting the layout of another open-source forum, I recommend esotalk.org/forum/, which I think has the best forum design that I've seen (but has too big fonts though). Or I could help copying esotalk's design, if it's a fit.
22:19:23	dom96	BitPuffin: nimrod c -r tests/async/tasyncawait
22:19:34	BitPuffin	fowl: heartbleed does not affect nimrod, we don't use ssl anyway lol
22:19:54	dom96	we do in babel :P
22:19:57	BitPuffin	dom96: too much work sorry
22:19:59	BitPuffin	jk
22:20:04	filwit	wan, yeah that forum looks slick
22:20:16	EXetoC	BitPuffin: not talking about that
22:20:31	EXetoC	it's an essential feature
22:20:32	BitPuffin	dom96: 5000
22:20:59	BitPuffin	EXetoC: babel doesn't count, we are not providing the service for babel :P
22:21:00	dom96	BitPuffin: cool, now write a kqueue implementation for the selectors module :P
22:21:03	BitPuffin	EXetoC: what?
22:21:09	BitPuffin	I meant dom96
22:21:12	BitPuffin	with the babel stuff
22:21:27	dom96	yeah, true.
22:21:30	fowl	BitPuffin, the next time you write packet serializing do it sloppily :D
22:21:43	EXetoC	BitPuffin: for the forum
22:21:47	BitPuffin	dom96: someone needs to get us an SSL cert for the forum login
22:21:47	dom96	I say we all voluntarily help the NSA.
22:22:02	BitPuffin	dom96: it's probably either you or Araq in that case
22:22:04	BitPuffin	probably Araq
22:22:15	BitPuffin	fowl: wat
22:22:15	dom96	Means they don't need to spend as much money on breaking our security, which means they use less of our tax dollars, which means we can pay less tax!
22:22:23	Araq	BitPuffin: let me work
22:22:27	BitPuffin	Araq: ?
22:22:34	Araq	don't highlight me
22:22:51	BitPuffin	don't look at highlights, rather
22:23:19	BitPuffin	plus it's not like it wasn't work related -_-
22:23:21	dom96	Araq: /msg *status detach #nimrod
22:23:52	BitPuffin	Araq: don't be rude
22:23:55	wan	dom96: Europeans don't pay taxes for the USA
22:24:03	BitPuffin	wan: at least we THINK
22:24:15	dom96	wan: Well we have our own NSA equivalents.
22:24:23	dom96	Gotta help 'em out too.
22:24:34	BitPuffin	GCHQ + FRA + NSA <3
22:24:35	fowl	BitPuffin, bleeding heart caused by lazyness https://github.com/fowlmouth/nimrod-enet/blob/master/pkt_tools.nim#L50
22:25:56	dom96	The one thing I hate about spring is all these damn insects flying around my room.
22:26:13	dom96	RIP moth.
22:27:11	BitPuffin	fowl: there is no heartbleed in there
22:27:21	wan	filwit: (home pic) the <pre> on the top of the page shouldn't use this font. The parenthesis look like square brackets, that's very confusing.
22:28:03	BitPuffin	fowl: well, if you wrapped malloc into a LIFO pool then maybe
22:28:06	fowl	BitPuffin, depends on how its used
22:28:16	filwit	wan, yeah that's an artifact of an older design i forgot to change.
22:28:18	dom96	haha: http://www.mumsnet.com/features/mumsnet-and-heartbleed-as-it-happened
22:28:25	filwit	wan, notice the code on the docs image doesn't have that font
22:28:44	BitPuffin	fowl: well the important thing about heartbleed was the stupid irresponsible fucking malloc wrapper
22:29:19	flaviu	filwit: It looks good, but the background is a bit too dark for my preference. I'm also not a big fan of the white in the syntax highlighting
22:29:24	filwit	wan, also things are not fully visualized in these designs. Like how the page content isn't all there, or the "lastest news"
22:29:31	fowl	BitPuffin, it would have been avoided with a simple check to see if the packet has enough bytes
22:29:36	BitPuffin	fowl: yes, the fact that it didn't check the incoming size was bad, but if they didn't wrap malloc then the extra bytes it would send back would not be anything useful, in fact it would probably crash for accessing memory it did not own
22:29:49	filwit	flaviu, okay, thanks for your input
22:30:00	BitPuffin	fowl: yes it would have, but the real source of the problem was malloc wrapper
22:30:02	Demos	The real problem in heartbleed was that it was disclosed in a really dangerous way
22:30:18	BitPuffin	Demos: it was really fucking strange how it was disclosed
22:30:38	BitPuffin	like some random cloud company I've never heard of before knew it one week before it was disclosed
22:30:53	BitPuffin	Amazon didn't even know before them
22:30:55	BitPuffin	like wtf
22:30:56	wan	filwit: by the way, what the hell happened with "Nimrod is efficient" kerning? `letter-spacing: calc(3px + random(10));`? The font kerning pairs informations for `ffi` seems to conflict with letter-spacing.
22:31:04	Demos	they should have waited until the patch was in wide circulation before telling the world, sure attackers could read the patch notes but still
22:31:26	BitPuffin	and Debian + FreeBSD security officers didn't even know until it was out there
22:32:03	filwit	wan, eh?
22:32:18	fowl	BitPuffin, someone slipped up and told a friend, they told a friend, etc etc, probably nobody wanted to be the public discloser of it
22:32:28	BitPuffin	yeah
22:32:33	BitPuffin	well it's fucked up anyway
22:32:37	flaviu	filwit: On the first link, web1
22:32:47	Demos	the other thing is that I dont think you could get specific info, you just got whatever was in memory, which makes the disclosure situation even more insane
22:32:59	flaviu	Demos: Someone got private keys
22:33:12	BitPuffin	Demos: well whatever was latest free'd
22:33:28	filwit	wan, flaviu, sorry i'm not sure what you're talking about. the ".efficient ..." thing on the homepage?
22:33:33	Demos	flaviu: right, but it was not "I want this user's keys" it was "I got the keys in memory"
22:33:47	filwit	wan, flaviu: if so, that's pretty much as it always has been...
22:33:50	wan	filwit: yes, the font in the docs image looks okay. Apart for, you know, italics. It's very weird for me to see italics in the code, maybe because I use vim in a terminal all day long, or maybe because italics shouldn't be used in code.
22:33:51	flaviu	filwit: No, in the body, under "Welcome to Nimrod"
22:34:15	fowl	BitPuffin, i bet there was a commit long ago patching malloc for openssl to fix the crash caused by the bug lol
22:34:47	filwit	wan, i disagree, i code with italics to distinquesh things with similar color but distinct differences (ei, properties vs methods)
22:34:52	Demos	so it was disclosed by cloudflare (I large CDN) in a public blog. They were like "look how on top of security we are"
22:35:16	filwit	wan, flaviu: oh, lol.. that's old content from the previous nimrod design
22:35:28	BitPuffin	fowl: haha, no the reason they wrapped malloc was for performance on some platforms
22:35:31	BitPuffin	which is a horrible reason
22:35:40	filwit	wan, flaviu: i just copy pasted it, as it wasn't really important for the design, just to get the idea across
22:36:03	wan	filwit: for italics, I'm just saying, I was surprised to see italics. Do other programming languages use italics in their <pre>?
22:36:18	flaviu	filwit: The point is that it looks like 'Nimrod is E ffi cent'
22:36:22	filwit	wan, italics are an IDE thing, not a language thing
22:36:30	filwit	wan, though idk if other sites use it
22:36:32	Demos	OK cloud flare posted the blog after the patch was out.... but only 1 hour after
22:36:47	wan	filwit: I'm talking about other languages' homepages or docs.
22:36:55	BitPuffin	http://article.gmane.org/gmane.os.openbsd.misc/211963
22:36:58	filwit	flaviu: yeah that's something odd with the font, or maybe i added a space on accident, dunno
22:36:58	*	brson quit (Ping timeout: 258 seconds)
22:37:08	filwit	wan, yeah i have no idea
22:37:26	filwit	wan, the exact way the code is highlighted is ultimately not up to me
22:37:40	wan	filwit: Is it reasonable to assume that italics are okay, or that italics are weird, because they are very often used / not very often used on docs pages around the web?
22:37:42	filwit	wan, and probably wont look exactly like my designs here
22:38:09	filwit	wan, that's an argument from popularity, and not a very good reason to make a choice :P
22:38:29	filwit	wan, but then, if most people are completely against them, i guess it makes sense not to have them
22:38:34	flaviu	I personally like the italics
22:38:38	filwit	me too
22:38:47	BitPuffin	fowl, Demos: Honestly it's really suspicious, this is not the kind of thing you accidentally do
22:40:20	EXetoC	fowl: want to review the pull?
22:40:37	EXetoC	and maybe someone else who is interested https://github.com/nimrod-code/opengl/pull/3
22:40:45	Demos	BitPuffin: who knows. I am going to assume it was some clueless MBA type who thought "oooh we could get PR", seems more likely than disclosing the bug (1h /after/ the patch came out) because of evil
22:42:32	BitPuffin	Demos: https://www.youtube.com/watch?feature=player_detailpage&v=X0_3rpA5_WE#t=3212
22:44:41	wan	filwit: yes, yes, I was trying to justify something based on its popularity, and that's not usually a good way to defend an argument. I don't remember seeing them around docs for python stuff or in <pre> in github repos' READMEs, and do find it highly unusual. Of course, each and everyone has their own experiences and usages, I'm just expressing my surprise.
22:45:27	BitPuffin	Demos: but yeah, OpenBSD is generally fantastic but the OpenSSL team is really weak imo
22:45:47	Demos	OpenSSL is not part of the OpenBSD project
22:45:55	Demos	in any way
22:46:46	filwit	wan, you're not allowed to express any emotion on the internet anymore. The NSA made that illegal last week. I'm going to have to report you to the authorities ;)
22:47:05	fowl	EXetoC, whats with the search in checkGlError()? just use $ error.TGlError
22:47:30	BitPuffin	Demos: oh it isn't?
22:47:36	BitPuffin	I always thought that
22:47:39	BitPuffin	guess not
22:47:53	BitPuffin	well thaaaaat explains everything
22:47:53	filwit	wan, just kidding of course, and thank you for all your feedback and interest
22:48:36	EXetoC	fowl: what about any unknown error codes? might be a little excessive though. anyway, I could use an 'in' check instead
22:48:45	filwit	wan, i just did what looks good to me in my designs, but it may be completely out of place and not something we want in the final production. You'll have to convince Araq of that though, not me.
22:48:59	fowl	EXetoC, they'll get something like (Invalid data!)
22:49:21	filwit	wan, btw, i haven't seen you around here before. are you new to Nimrod?
22:49:56	fowl	EXetoC, why change ptr GLchar to cstring? they should be the same thing
22:50:25	EXetoC	fowl: apart from the implicit casting
22:50:27	fowl	EXetoC, but the wrapper should reflect GL's man pages
22:50:29	*	brson joined #nimrod
22:50:30	BitPuffin	Demos: I was ordering plane tickets and stuff earlier and I was actually kinda worried about heartbleed
22:50:45	BitPuffin	like I dunno if Ryan air and stuff has patched their OpenSSL if they use it
22:50:48	BitPuffin	likely not
22:50:52	fowl	EXetoC, there is no implicit cast if GLchar is char then ptr GLchar is ptr char aka cstring
22:51:01	wan	filwit: I dropped in the irc some months ago, I have been lurking from time to time thanks to the irclogs mostly, I'm usually not talkative
22:52:37	filwit	wan, i see. well late-welcome i guess
22:52:58	EXetoC	fowl: cstring allows implicit casts from strings. I haven't found any functions yet that don't expect null-terminated strings
22:53:09	*	Varriount|Mobile quit (Remote host closed the connection)
22:53:11	wan	filwit: thanks
22:53:16	EXetoC	which is according to the convention, but I could have a closer look some other time just to be sure
22:53:24	*	Varriount|Mobile joined #nimrod
22:53:38	EXetoC	as it doesn't appear that often last time I checked
22:53:46	flaviu	BitPuffin: They use ASP.NET, so I don't think they're affected
22:54:10	BitPuffin	flaviu: ah
22:54:20	BitPuffin	flaviu: well I also ordered from uuuhh
22:54:24	BitPuffin	what are they called
22:54:43	BitPuffin	stanstedexpress.com
22:55:36	fowl	EXetoC, you'll still have implicit conv from strings
22:56:11	flaviu	BitPuffin: https://filippo.io/Heartbleed/#www.stanstedexpress.com
22:56:12	flaviu	That site didn't work for ryanair, their website had some sort of issue, but the end of the url was .aspx
22:56:29	fowl	oh
22:57:12	*	Varriount|Mobile quit (Client Quit)
22:57:19	BitPuffin	flaviu: yeah I basically just did the same thing but with a mcaffee tool
22:57:34	EXetoC	fowl: that was not the case yesterday
22:57:37	fowl	EXetoC, cstring used to be ptr char.. its not anymore
22:58:08	EXetoC	the current module uses it in some cases. it's binary compatible isn't it?
22:58:30	EXetoC	it uses cstringArray as well
22:59:32	BitPuffin	I would probably use NSS rather than OpenSSL nowadays
22:59:44	fowl	EXetoC, yes but weak to future opengl updates, make PGLchar* = cstring then, actually, when GLchar is char: type PGLchar* = cstring else: type PGLchar* = ptr GLchar
23:00:36	fowl	EXetoC, yea cstringarrays are annoying to deal with
23:01:50	EXetoC	I don't think they'll ever change
23:02:31	renesac	dom96, BitPuffin: http://www.openbsd.org/cgi-bin/cvsweb/src/lib/libssl/src/ssl/d1_both.c.diff?r1=1.7;r2=1.8;f=h <-- quality coding
23:03:10	fowl	EXetoC, you dont like that compromise?
23:03:27	BitPuffin	renesac: a lot better than openssl at least :P
23:03:37	EXetoC	fowl: it's fine
23:04:09	Araq	good night
23:04:11	EXetoC	I didn't include PGLchar etc though, because of the upcoming changes. should I? I don't think it'll break much code
23:04:48	filwit	i really wish the standard was just to use 'ref Foo' everywhere
23:04:52	fowl	EXetoC, not sure how you use this: GLeglImageOES* = distinct pointer, if you need to be able to isNil() on it, its easier to use `ptr object`
23:05:18	filwit	but that's because i'm still unclear on how to obtain a 'value type' of a 'ref object'
23:05:23	fowl	EXetoC, oh i see what u mean
23:05:48	fowl	filwit, you can dereference it
23:06:14	filwit	no, i mean when you need to pass the type for generics
23:06:22	renesac	and somehow, I think Microsoft already knew this bug for a long time...: https://github.com/joyent/node/commit/28c6e42e
23:08:04	EXetoC	fowl: with?
23:08:20	EXetoC	know about 'distinct with'? was added recently I think
23:08:20	fowl	EXetoC, you changed the types to GL*
23:08:48	fowl	i dunno what you're talking about
23:08:58	renesac	filwit, your bear looks too violent :P
23:09:03	EXetoC	fowl: "x = distinct with `isNil`" or something
23:09:11	renesac	but my complaint is only about the face
23:09:24	fowl	EXetoC, oh really?
23:09:58	fowl	EXetoC, i used regions in a recent wrapper
23:10:12	fowl	it looks nice: ptr[Lacewing, TTimer]
23:10:14	filwit	renesac, sorry brb
23:10:28	fowl	not really necessary at all though >_>
23:13:18	EXetoC	right
23:17:46	EXetoC	uh oh segfault "sigmatch.nim(506) maybeSkipDistinct"
23:17:57	BitPuffin	fowl: what was it you wanted me to test?
23:19:44	EXetoC	ok I'll just use 'ptr object' then
23:20:25	BitPuffin	EXetoC: pointer
23:20:44	EXetoC	fight!
23:21:37	fowl	BitPuffin, im trying to fix a bug before i have u run it.. its pissing me op >:/
23:23:12	BitPuffin	ah
23:23:18	BitPuffin	well let me know!
23:23:27	*	darkf joined #nimrod
23:32:47	filwit	renesac, back
23:33:04	filwit	renesac, that the mascot looks a bit too angry is the general feedback about the mascot
23:33:14	filwit	renesac, did you see the other designs?
23:33:28	renesac	the ones changing the color?
23:33:46	filwit	renesac, no, the new web/forum/docs designs
23:33:50	filwit	i'll PM you
23:33:57	renesac	not yet
23:36:17	*	zahary quit (Ping timeout: 250 seconds)
23:38:30	*	askatasuna quit (Quit: WeeChat 0.4.3)
23:43:55	NimBot	Araq/Nimrod devel 84c1aa9 Grzegorz Adam Hankiewicz [+0 ±1 -0]: Adds missing import.
23:43:55	NimBot	Araq/Nimrod devel 208ef29 Simon Hafner [+0 ±0 -0]: Merge pull request #1101 from gradha/pr_adds_missing_import... 2 more lines
23:44:27	NimBot	Araq/Nimrod devel de04c97 Grzegorz Adam Hankiewicz [+0 ±1 -0]: Improves startProcess docstring for poEvalCommand usage.
23:44:27	NimBot	Araq/Nimrod devel 6886a47 Simon Hafner [+0 ±1 -0]: Merge pull request #1102 from gradha/pr_improves_startProcess_docstring... 2 more lines
23:49:16	*	BitPuffin quit (Ping timeout: 240 seconds)
23:54:28	*	io2 quit (Quit: ...take irc away, what are you? genius, billionaire, playboy, philanthropist)
23:54:31	filwit	later folks
23:54:34	*	filwit quit (Quit: Leaving)
23:59:03	*	zahary joined #nimrod