00:00:00 --- log: started retro/09.12.26
00:01:00 * docl should get some sleep
00:01:04 <docl> 'night
09:32:19 --- join: erider (n=chatzill@unaffiliated/erider) joined #retro
10:01:01 <docl> hi erider
10:01:57 <erider> hi 
10:16:14 * docl has been debating cryonics
10:17:18 <docl> and programming retroforth too :)
10:39:48 <erider> brb
10:39:53 --- quit: erider ("ChatZilla 0.9.86 [Firefox 3.5.6/20091201220228]")
11:03:38 --- quit: virl (Remote closed the connection)
12:48:03 <docl> : sav ` push ` r ; immediate
12:48:26 <docl> more optimal: : sav ` dup ` push ; compile-only
12:48:47 <docl> but the question is, is sav really a good name for the word?
12:49:02 <docl> maybe "pr" for "push r" or something
13:24:17 --- quit: docl (Remote closed the connection)
13:24:32 --- join: docl (n=luke@67-60-48-82.cpe.cableone.net) joined #retro
13:25:04 <docl> http://gist.github.com/264046
13:25:46 <docl> I'm not sure why, but when I try to compile calls to the commands "+ @" after an eval it messes up.
13:29:52 <docl> I must have messed up the return stack somehow
13:42:07 <docl> of course I can simplify the word to get around such issues, e.g. : access ( n"- ) ' + @ ;
13:52:12 <docl> here's a simpler example:
13:52:14 <docl> : test s" 10 . 20 ." dup getLength eval words ; test
16:06:30 <crc> docl: the output is missing due to remapping of "emit" by the "eval" function
16:17:23 <docl> oh yeah, that makes sense
16:17:33 <crc> I should document that
16:18:15 <crc> or maybe change eval to revector accept
16:25:48 <crc> docl: try this eval function: http://gist.github.com/264093
16:28:53 <docl> hmm.
16:29:00 <docl> : access ( n"- ) 32 accept tib dup getLength eval + @ ;
16:29:00 <docl> create mylist 1 , 2 ,
16:29:00 <docl> 1 access mylist
16:30:25 <docl> tos: 7293 nos: 2486
16:33:09 <docl> evidently 2486 is the number stored at memory address 1
16:33:53 <docl> perhaps the addition is being delayed until after the @ somehow?
16:39:22 * crc is trying to find out
16:42:38 <crc> generating an execution trace at present
16:43:31 <crc> 67mb trace generated :)
16:52:43 <crc> docl: why are you using eval here? (just out of curiosity)
16:56:39 <docl> well I'm not anymore, since I figured out that ' does the job in this context
16:56:43 <docl> : access ( n"- ) ' + @ ;
16:59:46 <docl> I seem to recall "wsparse eval" used to work in the older versions of retro.
16:59:52 <crc> imo, it's best not to use eval unless there's no other way
16:59:59 <crc> docl: yes it did
17:00:24 <crc> the new eval is likely more limited, and definitely less well-tested than the one in older retro systems
17:01:00 <docl> you just added it recently right?
17:01:06 <crc> yes
17:03:11 <retro-commit> [crcx/retro10] 261ca3: cleaner "accept" in core; added "ekey" (replaces "...
17:04:09 <crc> since this eval is substantially different than the old one in how it works, it may break/alter things
17:04:57 <docl> understood.
17:05:16 <crc> the old one was actually a duplicate of the interpret loop, using strings as input
17:05:36 <crc> this one remaps the 'ekey' word used by 'accept'
17:06:02 <crc> the devectoring may be causing issues with code following the eval
17:11:57 <docl> hmm. I can't figure out why + or @ would be affected by key or emit
17:12:25 <docl> and those devectorings happen at interpret time, not compile time...
17:21:40 --- quit: yiyus (farmer.freenode.net irc.freenode.net)
17:21:40 <crc> the 'eval' doesn't happen until the word exits
17:22:05 --- join: yiyus (i=12427124@je.je.je) joined #retro
17:23:55 <crc> docl: try this to see what I mean:
17:23:56 <crc> : access ( n"- ) 32 accept tib dup getLength eval words ;
17:23:57 <crc> create mylist 1 , 2 ,
17:24:15 <crc> err, let me paste again
17:24:25 <crc> : a cr ." hello" cr ;
17:24:25 <crc> : access ( n"- ) 32 accept tib dup getLength eval words ;
17:24:25 <crc> create mylist 1 , 2 ,
17:24:25 <crc> 1 access a
17:26:15 <docl> hmm. it never gets to "words"
17:26:47 <crc> docl: it does with the newest eval (pull from my retro10 repo and build a fresh image)
17:27:39 <crc> docl: words doesn't show any output with the old image, since 'emit' is changed to a non-op in that 'eval'
17:28:25 <crc> docl: the new one has been restructured to display output properly
17:29:52 <docl> ah ok
17:30:19 <docl> : test1  @ + ;
17:30:19 <docl> : test2 ( n"- ) 32 accept tib dup getLength eval ;
17:30:19 <docl> : access test2 test1 ;
17:30:19 <docl> create mylist 1 , 2 ,
17:30:21 <docl> 0 access mylist .
17:30:27 <docl> still doesn't work
17:30:58 <crc> docl: the eval still doesn't happen until the calling word exits
17:31:17 <crc> interpreter calls word, word (or something it calls) invokes eval, which remaps input
17:31:19 <docl> so the word that calls from the interpreter, in other words?
17:31:28 <crc> new input isn't processed until the interpreter gets control again
17:31:48 <docl> oooh ok
17:38:48 <crc> docl: see if this is clear enough: http://gist.github.com/264114
17:41:43 <docl> looks pretty clear.
17:44:56 <retro-commit> [crcx/retro10] f9e1dc: add notes on "eval"
17:45:34 <crc> I'm probably going to start dividing the documentation into smaller files, to make it easier to manage
17:53:25 <docl> here's something that works:
17:53:26 <docl> : prepend ( $$-$ ) here -rot 2 for dup here swap getLength dup allot copy next 0 , ;
17:53:29 <docl> : access ( n"- ) 32 accept tib  s"  + @ ." swap prepend dup getLength eval ;
17:53:32 <docl> create mylist 1 , 2 ,
17:53:34 <docl> 0 access mylist
17:53:57 <retro-commit> [crcx/forthlets] 4c4d47: use ijkl in console game demos
17:58:19 <crc> seems like overkill though
17:58:40 <docl> yeah, but it proves it can be done that way :P
17:58:55 <crc> true
17:59:01 <docl> : access ' + @ ; is the simple way
17:59:08 <crc> yup
17:59:36 <docl> but I'm thinking for some kinds of scripting, it might be more convenient to be able to use text.
18:00:22 <crc> true
18:00:30 <docl> in which case you want something like prepend
18:00:34 <crc> some macros could go a long way towards making it readable
18:02:21 <docl> something like "code { }" maybe?
18:03:30 <crc> that could work nicely
18:05:25 <crc> bbiab, dinner time 
18:05:28 <docl> ok
19:11:04 <docl> http://gist.github.com/264138
19:21:57 <retro-commit> [crcx/retro10] 3a6a6d: add ifNotExists
19:22:53 <crc> docl: looks good
19:23:09 <crc> with ifNotExists, you can do:
19:23:11 <crc> ifNotExists prepend { : prepend ( $$-$ ) here -rot 2 for dup here swap getLength dup allot copy next 0 , ; }
19:23:25 <crc> to only define prepend if the string library isn't loaded
19:48:56 <retro-commit> [crcx/retro10] 94f699: add some more rst files to docs/notes
20:03:05 <docl> ooh, nice
20:06:52 <crc> btw, I've updated my online demo at http://crcx.ath.cx:52850/ with a toggle on the html filter and more of the forthlets
20:10:27 * docl is taking a look
20:13:01 <docl> is the chorder loaded?
20:13:12 <crc> yes
20:13:27 <crc> wait a second...
20:14:19 <crc> now it is
20:14:34 * crc forgot to change the ifExists line to the new syntax in it, sorry 
20:14:39 <docl> ok :)
20:19:13 <docl> nice
20:19:38 <docl> is it still running on a canvas? it seems faster, but that could be my imagination
20:19:56 <retro-commit> [crcx/forthlets] 6e8080: new ifExists syntax
20:20:02 <crc> the text is no longer drawn to the canvas, just layered over it
20:20:39 <docl> cool
20:20:55 <docl> it still takes some time to render though. 
20:21:05 <crc> I've still not figured out how to deal with the scrolling issues...
20:21:28 <crc> docl: yes it does. I'm still evaluating options for improving display speed :(
20:22:01 <docl> would it be possible to just update the changed areas?
20:25:19 <crc> I have no way of tracking changes yet
20:26:24 <docl> would a position-specific emit be possible, I wonder?
20:26:36 <docl> pos-emit
20:26:47 <docl> ( xyc-" )
20:26:50 <crc> yes, but it'd have to be tailored for each vm/host os
20:27:15 <docl> yeah I can see how that would be a pain.
20:38:23 <docl> well, using ifExists we can probably write fairly portable code even so.
20:38:56 <crc> even that won't help much with system-dependencies though :(
20:47:15 <docl> does the sdl version have the layering improvement? or is that just javascript?
20:47:24 <crc> just javascript
20:47:33 <crc> sdl is *brutally* slow at present
20:48:24 <docl> it seems to take as long to print blank spaces as it does regular characters
20:49:13 <crc> yup.
20:49:43 <crc> the font drawing could be improved a lot (use of clipping, bitblt, etc) but I've not had a lot of time to play with it
20:52:16 * docl will look at it some more
20:58:01 * crc seldom runs the sdl port anymore
21:15:45 <docl> hmm, if you set the window height to 150 you can still see the entire block editor and it renders faster. still impossible to type in rem though since it does not accept input during refresh.
21:25:02 <docl> using red with it is not so bad.
21:30:44 <crc> docl: load this and save:
21:30:47 <crc>  fastdraw ;
21:30:47 <crc> here is v ] ['] fastdraw ['] redraw :is (v) ['] redraw :devector ;
21:30:54 <crc> see if that helps
21:37:28 <docl> speeds it up tremendously!
21:37:46 <docl> not in rem though, at least not the way I loaded it.
21:38:57 <retro-commit> [crcx/retro10] fbcf1f: add "fast-render" to stage3
21:40:48 <crc> (v) now uses this and I'm starting to add it to the forthlets
21:41:02 <docl> fast-render seems to break the console version though :(
21:41:20 <crc> odd
21:41:24 * crc is checking now
21:43:10 <crc> docl: the console stuff still works for me
21:43:50 <docl> it's weird. I just see the bottom bar and the status bar.
21:44:00 <crc> run new again
21:44:11 <docl> ahh :)
21:44:16 <docl> works
21:44:34 <docl> why would it do that?
21:45:02 <crc> the memory is likely filled with ascii 0, which displays nothing
21:49:15 <docl> hmm, rem now seems to have been broken, it waits until you exit the key loop by pressing z before displaying the screen.
21:49:23 <docl> (in sdl)
21:50:41 <docl> so what you end up seeing is a blank screen
21:51:11 <crc> I just pushed a fix to fast-render
21:51:57 <retro-commit> [crcx/retro10] 0a4267: fix to fast-render
21:53:27 <docl> cool!
21:54:42 <crc> it should work ok now
21:56:42 <docl> works great!
21:57:56 <retro-commit> [crcx/retro10] 628c4e: move "words" to stage3; use "fast-render" in "word...
22:00:56 <retro-commit> [crcx/retro10] 8b3f24: tweak to "see" with fast-render
22:02:00 <retro-commit> [crcx/retro10] b9f73c: drop fast-render from see; didnt like the results
22:15:04 <retro-commit> [crcx/forthlets] 39951a: fast-render in console games
22:16:15 <docl> does "forget" unallot heap space?
22:16:41 <crc> yes
22:17:19 <docl> nice
22:51:50 <docl> : :screen  (block) create here 512 copy 512 allot 0 , ;
22:55:46 <docl> : screen:  (block) over getLength copy ;
23:01:56 <retro-commit> [crcx/retro10] c9ae49: reformat to rst
23:08:05 <retro-commit> [crcx/retro10] 305480: LICENSE now rst format
23:21:17 <crc> docl: stack comments please ;)
23:29:38 <docl> ok
23:30:30 <docl> : :screen  ( "- ) (block) create here 512 copy 512 allot 0 , x ;
23:30:30 <docl> : screen:  ( a- ) (block) over getLength copy ;
23:31:24 <docl> example:
23:31:24 <docl> :screen temp n temp screen: forget temp
23:31:44 <docl> the names don't feel quite right
23:32:56 <docl> maybe block-as ( "- ) and as-block ( a- )
23:38:53 <docl> new
23:38:57 <docl> 0 i : block-as ( "- ) (block)
23:38:57 <docl> 1 i   create here 512 copy 512 allot 0 , x ;
23:38:57 <docl> 2 i : as-block ( a- ) (block) over getLength copy v ;
23:38:57 <docl> e 
23:38:57 <docl> block-as temp n temp as-block forget temp
23:59:59 --- log: ended retro/09.12.26