<<18-05-2013>>

---

01:18:33	*	fowl quit (Ping timeout: 256 seconds)
01:30:08	*	fowl joined #nimrod
01:39:30	*	q66 quit (Remote host closed the connection)
04:19:39	reactormonk	fancy, so depending on type, stuff is written to the RAM or not
04:19:50	reactormonk	any idea how to represent that in nimrod?
04:20:14	fowl	example?
04:20:45	reactormonk	#define __ATTR_PROGMEM__ __attribute__((__progmem__))
04:20:53	reactormonk	#define PROGMEM __ATTR_PROGMEM__
04:20:57	reactormonk	typedef int8_t prog_int8_t PROGMEM;
04:22:22	reactormonk	so prog_int8_t is stored in another address space
04:23:14	fowl	you should just import prog_int8_t like type prog_int8_t* {.importc.} = int8
04:23:49	fowl	er wait you can only do that with {.header.}
04:24:12	fowl	i dunno, didnt araq's extern: "$1 PROGMEM" work
04:24:21	reactormonk	didn't try
04:37:27	fowl	you have to link with a DLL?
05:02:38	reactormonk	nope, only code
05:02:45	reactormonk	install arduino
05:03:08	reactormonk	fowl, http://arduino.cc/en/Reference/HomePage
05:04:19	fowl	ok then you should be able to do type prog_int8_t* {.importc, header: "the/header.h".} = int8
05:05:27	fowl	you should test it with var x{.exportc: "FINDME".} : prog_int8_t and make sure its declared as prog_int8_t in the output.c
05:20:01	*	OrionPK quit (Quit: Leaving)
05:23:00	reactormonk	doesn't need to be... maybe
05:24:07	reactormonk	not needed, people can just use the progmem pragma
05:24:10	reactormonk	{.pragma progmem, extern: "PROGMEM $1".}
05:24:42	fowl	for what
05:24:49	reactormonk	the only slight problem might be const uint16_t PROGMEM port_to_mode_PGM[] = {
05:24:52	reactormonk	NOT_A_PORT,
05:24:54	reactormonk	NOT_A_PORT,
05:24:56	reactormonk	(uint16_t) &DDRB,
05:24:58	reactormonk	(uint16_t) &DDRC,
05:25:00	reactormonk	(uint16_t) &DDRD,
05:25:02	reactormonk	};
05:25:04	reactormonk	... for storing stuff in the PROGMEM
05:25:24	fowl	that type decl has PROGMEM in it, surely you dont need to repeat that when you declare variables of that type..
05:29:17	reactormonk	# const uint16_t PROGMEM port_to_mode_PGM[] = {
05:29:19	reactormonk	goes to?
05:29:30	reactormonk	ehm, does nimrod have an int16?
05:29:49	fowl	yea
05:29:52	fowl	int16, uint16
05:30:24	fowl	var port_to_mode* {.progmem, importc.}: ptr array[0.. <1024, uint16]
05:32:08	reactormonk	what's the array constructor syntax? { } or [ ]?
05:34:09	fowl	[] but you dont want that
05:34:37	reactormonk	but?
05:35:46	fowl	that data already exists, you just need access to it from nimrod
05:36:50	reactormonk	const port_to_output_PGM* {.progmem, importc.}: ptr array[0..4, uint16]
05:38:41	reactormonk	and I assume I gotta add the header to every importc around?
05:39:59	fowl	probably
05:40:03	fowl	i dont think you want const there though
05:40:49	fowl	also 4 is the highest bounds for your hardware only
05:41:44	reactormonk	so what
05:41:51	reactormonk	that's generally arduinos
05:44:39	reactormonk	when defined(__AVR_ATmega8__):
05:44:44	reactormonk	now that doesn't work ^^
05:45:40	fowl	so your arduino wrapper will be specific to your hardware
05:45:46	reactormonk	pins_arduino.nim(25, 2) Error: invalid pragma: {.pragma, progmem, extern: "PROGMEM $1".}
05:45:53	fowl	for no real reason
05:45:56	reactormonk	+/-
05:46:14	reactormonk	I guess I'll importc that one too
05:50:21	*	xcombelle joined #nimrod
05:53:14	reactormonk	any idea about the pragma?
05:53:43	reactormonk	{.pragma: "progmem", extern: "PROGMEM $1".}
05:53:46	reactormonk	doesn't work either
05:53:53	fowl	no, sorry
05:55:10	fowl	try
05:55:25	fowl	{.pragma: progmem, extern:"PROGMEM $1".}
05:55:46	reactormonk	sweet
05:56:04	fowl	yea i have to look it up every time i use it <_<
05:56:45	reactormonk	so it picks up my nimrod.cfg that tells it not to compile :-)
05:57:56	reactormonk	any way I can tell the compiler where to stuff the files?
05:58:46	reactormonk	how do you deal with http://sprunge.us/BQHV
05:58:48	fowl	i dont follow, stuff what files
05:58:56	reactormonk	the *.c files
05:59:28	fowl	the generated .c files?
06:00:15	fowl	reactormonk: use importc/header for DEFAULT, EXTERNAL and INTERNAL
06:01:25	reactormonk	Error: internal error: getTypeDescAux(tyExpr)
06:03:04	reactormonk	maybe related to https://github.com/Araq/Nimrod/issues/147
06:05:56	reactormonk	oh
06:06:01	reactormonk	var DEFAULT*: expr {.importc, header: "Arduino.h"}
06:06:03	reactormonk	guess why
06:06:39	reactormonk	#define DEFAULT 0
06:06:43	reactormonk	what's this type?
06:06:46	reactormonk	int?
06:06:58	fowl	cint
06:07:20	fowl	c's int is int32, nimrod's is int64 if you have it, cint is c's int
06:08:22	reactormonk	var
06:08:24	reactormonk	DEFAULT*: cint {.importc, header: "Arduino.h"}
06:08:31	reactormonk	any reason why this gives me 'invalid indent'?
06:08:48	reactormonk	http://sprunge.us/gabM
06:09:35	fowl	oh
06:09:42	fowl	the pragma goes after the visibility
06:09:48	fowl	var x* {..}: ty
06:12:13	reactormonk	but the other way round with procs
06:13:26	reactormonk	for stuff like int digitalRead(uint8_t);
06:13:35	reactormonk	grab the corresponding C file or simply importc?
06:27:57	reactormonk	what was the 'no mangle' again?
06:28:08	reactormonk	I need to define setup and loop accordingly
06:39:17	*	fowl quit (Ping timeout: 256 seconds)
06:51:09	*	fowl joined #nimrod
07:00:40	Araq	reactormonk: the 'extern', importc or exportc means "no mangle"
08:05:31	NimBot	Araq/Nimrod a9f3450 Araq [+0 ±3 -0]: preparations for the new symbol binding rules for clean templates
08:19:23	*	zahary_ joined #nimrod
08:32:13	NimBot	Araq/Nimrod 89284c3 Araq [+0 ±2 -0]: fixes #440
08:32:14	NimBot	Araq/Nimrod 20007dc Araq [+0 ±1 -0]: fixes #437, fixes #436
09:14:58	*	xcombelle quit (Read error: Connection reset by peer)
09:56:37	NimBot	Araq/Nimrod 057f83b Araq [+0 ±1 -0]: fixes #399
09:56:37	NimBot	Araq/Nimrod 87cc542 Araq [+0 ±1 -0]: fixes #119
11:01:59	*	q66 joined #nimrod
11:08:08	*	Endy joined #nimrod
11:21:07	*	[1]Endy joined #nimrod
11:23:24	*	Endy quit (Ping timeout: 245 seconds)
11:25:28	*	[1]Endy quit (Ping timeout: 246 seconds)
12:23:33	*	Endy joined #nimrod
13:03:29	*	fowl quit (Ping timeout: 256 seconds)
15:55:31	*	OrionPK joined #nimrod
16:07:28	*	Trix[a]r_za is now known as Trixar_za
16:23:31	reactormonk	Araq, so extern: "loop" if I want to define "loop"
16:44:33	Araq	reactormonk: well better use 'importc' or 'exportc'
16:57:18	reactormonk	Araq, can I tell nimrod to generate the .c files in the same dir as the .nim files?
16:59:10	Araq	there is --nimcache:DIR but I never tested to use the same dir ;-)
17:01:05	reactormonk	hm. does --nimcache:DIR depend on $PWD or the nimrod.cfg file?
17:01:32	Araq	no
17:01:45	Araq	(whatever that means)
17:02:24	reactormonk	nope, you can't specify "."
17:02:38	reactormonk	... at least in the config file
17:04:01	Araq	you can use --nimcache:$projectdir I think
17:04:28	reactormonk	oh fuck, that's gonna be ugly
17:05:09	reactormonk	because ino (the arduino build tool) does its own dependency scanning - and I assume nimrod doesn't bother too much about that in the generated source?
17:05:33	Araq	just don't use ino then
17:05:54	Araq	the less you use the more likely it ends up working
17:07:11	reactormonk	the less I use, the more I have to build myself
17:07:33	Araq	yeah and then you can actually fix things
17:08:53	reactormonk	not interested in developing a serial iterface for nimrod
17:09:40	reactormonk	hmm, I don't need that.
17:10:35	Araq	you can also use --genscript to make nimrod generate a shell script that does the compilation and then edit that
17:11:15	reactormonk	there's a separate command 'upload' which handles the serial communication for me
17:33:18	reactormonk	any way I can symlink the nimbase.h into my dir?
17:33:40	Araq	I'd copy it over instead, it doesn't change anyway
17:36:34	reactormonk	http://sprunge.us/VOYB huh
17:37:47	reactormonk	http://sprunge.us/djXU the whole think
17:37:54	*	Endy quit (Ping timeout: 264 seconds)
17:38:46	Araq	who is general failure and why is he reading my hard disk?
17:39:06	reactormonk	yep.
17:43:18	reactormonk	now that's gonna be ugly :-/
17:43:50	reactormonk	because if I do the compilation myself, I have to do all the ugly shit with -D__ARDUINO_MX87239__
17:53:30	reactormonk	Araq, can I unlink signal.h? Arduino doesn't suppor that part
17:54:34	Araq	-d:noSignalHandler may work
17:55:06	Araq	but ugh it's in nimbase.h ...
17:55:16	Araq	we shouldn't do that
18:01:06	NimBot	Araq/Nimrod 75a0bb1 Dominik Picheta [+12 ±6 -1]: Implemented new Nimrod website designed by @PhilipWitte.
18:01:06	NimBot	Araq/Nimrod 041efeb Dominik Picheta [+0 ±2 -0]: Website adjustments; added table style and span.pre style.
18:01:40	reactormonk	is the new website online by now?
18:01:50	dom96	Not yet
18:02:17	reactormonk	aww
18:06:41	reactormonk	src/nimcache/nimbase.h:485: error: size of array ‘assert_numbits’ is negative
18:07:48	reactormonk	Araq, http://sprunge.us/KcLQ
18:07:55	reactormonk	src/nimcache/system.c:6046: error: ‘SIGSEGV’ undeclared (first use in this function)
18:08:57	reactormonk	forgot to rebuild the sketch. http://sprunge.us/BifB
18:09:53	reactormonk	yep, and dlfcn.h doesn't exist on an arduino for sure.
18:13:54	reactormonk	not sure about sys/mman.h either
18:17:05	reactormonk	oh, and you can load stuff into the flash (which is way bigger than the RAM) if you don't need to modify it. But apparently no dynamic allocation.
18:17:30	reactormonk	what goes down the train without dynamic allocation? Objects? Or can I allocate them statically.
18:17:44	Araq	sure you can
18:18:52	reactormonk	can I exclude named .h files?
18:20:40	reactormonk	aka make --d:noSignalHandler exclude .h and --d:noDynamicAlloc
18:30:29	*	bloup joined #nimrod
18:30:33	*	bloup left #nimrod ("WeeChat 0.3.7")
18:45:30	*	Endy joined #nimrod
19:04:26	*	fowl joined #nimrod
19:11:19	*	Trixar_za is now known as Trix[a]r_za
19:18:49	*	Trix[a]r_za is now known as Trixar_za
19:29:23	dom96	fowl: Could you make a pull request for your dsl module? (and clean it up if needs be)
19:29:47	dom96	We're releasing 0.9.2 quite soonish ;)
19:36:25	fowl	ok
19:37:43	*	exhu joined #nimrod
19:37:45	reactormonk	Araq, should I try wrapping a bunch of when(not(noSignalHandler)) and same for dynamic alloc?
19:37:56	reactormonk	gotta say though I don't even know the syntax for staic alloc...
19:38:11	reactormonk	wait, just tell it not to make it a reference?
19:47:00	Araq	reactormonk: yes.
19:47:18	Araq	hi exhu ... the macros bug you found it actually quite hard to fix :P
19:47:37	reactormonk	Araq, is it possible to make the dynamic alloc optional?
19:47:39	Araq	switching to luajit would fix it though ... ;-)
19:47:53	exhu	hi Araq, i'm glad you fixed it quickly, thanks!
19:49:00	exhu	Araq, will luajit be a dependency for a produced program, or only during compilation phase?
19:49:17	reactormonk	exhu, I assume the name 'jit' implies a runtime dependency
19:49:57	Araq	no, only for compilation
19:50:20	Araq	exhu: I didn't fix it ...
19:51:02	exhu	Araq, eh, i've post several issues recently, the one that prevented compilation of unit tests is fixed.
19:51:24	Araq	oh yeah ... that was easy and reported 3x
19:52:26	exhu	Araq, and the one you can't fix, i don't remember which :) i stopped ditching on macros, no ORM etc stuff for now.
19:52:41	*	Endy quit (Ping timeout: 252 seconds)
19:53:31	Araq	well I know how it can be fixed but it's too risky, we want to release 0.9.2 tomorrow ...
19:53:38	fowl	exhu: its probably easier to do what you by passing the information directly like tablemacro(tablename, tuple[field1: int, field2: string]) and you should be able to put pragmas in the tuple ty like you wanted
19:56:55	exhu	Araq, and reading the antlr site, i've spotted a nice phrase by the authour of antlr tool (scanner-parser-ast parser generator), "Why program by hand in five days what you can spend twenty-five years of your life automating?" =) So i think i'll do it by hand rather than develop macros for the rest of the year -)
19:59:55	Araq	the macro ain't for you ... it's for your users :P
20:00:21	Araq	plus an ORM screams for macros
20:00:48	reactormonk	isn't the NoSQL movement strong nowadays ;-)
20:01:13	exhu	i'll probably leave macro fun for v2.0 of the program/lib i'm developing, i desperately need a working program that does the work, then i'll be ready to spend more time on improvement and unification.
20:02:16	exhu	the second urging feature for my app is the gui toolkit, don't know which way to go here.
20:02:46	reactormonk	Araq, or do we need a target "microcontroller" ;-)
20:04:08	exhu	i've studied some xlibs/xcb stuff, it's a real pain, so many legacy things, mixture of new and old xlib and xcb, no docs of how to use it etc. Win32 API is much better in this regard.
20:05:20	exhu	i remember writing simple wrappers for win32 api gui in the 90s -- early 2000s, it was a piece of cake.
20:05:26	Araq	reactormonk: we have that target already
20:05:34	reactormonk	Araq, tell me
20:08:28	Araq	reactormonk: http://nimrod-code.org/nimrodc.html#nimrod-for-embedded-systems
20:08:54	Araq	I know --gc:none doesn't work with -d:useMalloc anymore, will fix it tonight
20:09:19	reactormonk	Araq, argh. Not enough googability on nimrod :-(
20:11:40	reactormonk	src/nimcache/nimbase.h:256:20: error: signal.h: No such file or directory
20:11:43	reactormonk	still.
20:11:52	reactormonk	there are no signals when there is no OS.
20:15:18	Araq	yeah fix it, nimbase.h doesn't need it anyway
20:18:07	reactormonk	system.c still includes it though
20:20:05	Araq	even with -d:noSignalHandler ?
20:20:10	reactormonk	yep
20:20:23	reactormonk	d = noSignalHandler
20:20:28	reactormonk	if this works in nimrod.cfg
20:22:03	Araq	it should
20:22:10	Araq	well fix it then :P
20:22:50	reactormonk	ok
20:23:00	reactormonk	where should I start looking?
20:23:22	Araq	lib/system/ansi_c.nim
20:23:27	reactormonk	kk
20:31:32	fowl	i think macro_dsl needs a new name
20:31:40	fowl	its less a dsl and more a collection of helpers
20:31:54	Araq	just put it into macros.nim
20:32:08	Araq	but leave out the controversial parts :P
20:32:27	Araq	not use we need := to generate an assignment for instance
20:32:31	Araq	*not sure
20:33:09	dom96	Does anyone have an example of how to use macros.quote?
20:34:18	fowl	dom96: quote: let x = 42
20:35:22	dom96	How do you interpolate?
20:35:39	fowl	Araq: its <- now, and it only matches on PNimrodNodes :p
20:36:09	fowl	dom96: quote: let x = `y`
20:36:33	fowl	iirc it resolves symbols too early to be meaningful in a macro but ill try it real quick and see
20:36:48	Araq	yeah it's buggy/wrong
20:37:11	*	dom96 wonders why the example in the docs shows the usage together with a 'do'
20:37:17	fowl	thats quote do: my bad
20:37:33	dom96	ahh, so you do need 'do'
20:38:03	fowl	dom96: i've been using parseExpr() a lot lately
20:38:32	fowl	newStmtList(parseStmt("let x = $1" % $var"), ...)
20:38:35	dom96	fowl: Yeah, that's what I used too. But I feel that 'quote' may be better.
20:38:57	dom96	Also if it's broken it should be fixed before we release ;)
20:39:22	Araq	yeah come up with a good example and I'll fix it
20:41:42	fowl	https://gist.github.com/fowlmouth/5605720
20:53:20	dom96	It's definitely not good that Eurovision is happening today...
20:56:16	*	exhu quit (Quit: Ex-Chat)
20:58:19	NimBot	Araq/Nimrod 742f9fe Simon Hafner [+0 ±1 -0]: nimbase.h doesn't need signal.h
21:00:04	fowl	https://gist.github.com/fowlmouth/f669d9722c72483a7608#file-macro_dsl-nim-L205
21:00:11	fowl	are those names good
21:00:20	fowl	(and the rest of the funcs above)
21:11:08	Araq	newBlockStmt*(label: PNimrodNode = nil, statements: varargs[PNimrodNode] = []): PNimrodNode
21:11:23	Araq	that doesn't work properly I think
21:14:10	fowl	how about a separate newBlockStmtLabelled(label: PNimrodNode; statements: varargs[PNimrodNode] = [])
21:15:49	Araq	in fact I dislike your heavy usage of varags
21:15:55	Araq	what's wrong with openarray?
21:16:06	Araq	enforces the [] which make it clearer imo
21:26:40	fowl	i updated it
21:29:38	Araq	copyChildrenTo -> addChildrenTo
21:30:15	Araq	oh hrm
21:30:20	Araq	you do copy
21:32:22	Araq	fowl: ever used 'body'?
21:32:43	Araq	can't think of an example where you don't know whether it's a proc or a block...
21:34:37	Araq	template first*(n: PNimrodNode; cond: expr): PNimrodNode {.immediate, dirty.}
21:34:39	Araq	-->
21:34:56	Araq	template findIt*(n: PNimrodNode; cond: expr): PNimrodNode {.dirty.}
21:35:34	Araq	and it should be recursive :P
21:38:02	Araq	or make it 'findChild'
21:38:41	Araq	has_argument_named should be hasArgOfName I think
21:41:54	fowl	Araq: yea i use body() all the time so i dont have to remember the stmt list is in the 7th slot for procs and the 2nd for blocks
21:42:29	Araq	yeah ok but do you need the polymorphism?
21:42:47	Araq	I'd define body simply for procs
21:43:39	fowl	it should be useful to get the body out of anything that has a stmt list
21:44:47	Araq	meh fine
21:45:01	Araq	I still can't see why I ever would want that for blocks but alright
21:45:26	Araq	I mean why I would want that for 'block' but not also for 'while' for instance
21:47:33	fowl	because i hadnt needed it for while yet :p
21:48:18	Araq	suprisingly you needed it for block :P
21:54:39	fowl	dunno how to make findIt recursive
21:54:52	fowl	does that even make sense for a template
21:58:45	Araq	no you need an aux proc to get it recursive
21:58:54	Araq	but it's fine just name it 'findChild'
21:59:07	Araq	that suggests it's not recursive I hope
22:24:24	fowl	how about proc add_ident_if_absent*
22:24:51	Araq	make it camelCase and it's fine I think
22:30:35	fowl	ok im gonna walk to the donut shop and so if theres anything else (i just updated it) ill fix when i get back
22:30:58	Araq	well more docs would be nice
22:31:16	Araq	but I'll likely include it to macros.nim anyway