XLua, MMF2 Extension

so uh sometime tomorrow or wednesday i'll have to send you something that's bugging me because i don't know if it's something that me, xlua, or mmf is doing

in short, something is making objects invisible; they're apparently in the right positions since they paste fine but they don't actually show up anywhere (and i'm p. sure i'm not changing their transparency or hiding them)

but i'll have to tear down the engine first and that'll take a while

also the 'open jit' action displays as 'open bit' on the grid so i dunno whether it's loading the wrong library or just showing the wrong name

triple post!

also is it just me or do the mmf.Object.Overlaps*() functions not like loops

object.pushFromBackground = function (input_obj) ---[[ local BREAK = 0 -- unf WHYYY local obs = true while obs do BREAK = BREAK + 1 -- temp input_obj:setY(input_obj.y - 1) object.saveData(input_obj) obs = mmf.Object.OverlapsBackground(input_obj.sysid) if BREAK > 100 then --or not(mmf.Object.OverlapsBackground(input_obj.sysid)) then break end end--]] end

object.saveData() basically just does mmf.Object.SetY() and mmf.Object.Update() (and others too obviously), so no i'm not forgetting to update the object during the loop

however this loop keeps running until BREAK is greater than 100 (the negated overlap function does jack when it's uncommented), even when the object is no longer touching a backdrop... what

This post has been edited by Xgoff on Jun 26 2009, 10:05 PM

isn't the "while obs do" an infinite loop seeing as 'while' keeps looping until it is broken

the idea is that once mmf.Object.OverlapsBackground() returns false it'll break

but it doesn't because i think something's not playing nice

Something you need to be aware of Update() is that it just sets a flag. No matter how many times you call it, it won't be acted on until the end of the event loop.

As far as Object.OverlapsBackground(), it may be tracking updated coordinates for sprites separately and then updating at the end of the loop. Whip me up a test MFA exhibiting the problem and I'll try using different values.

http://file.walagata.com/w/supermetalmario/slowcoll.zip

there should be a way to do this since it's possible with fastloops, and those don't wait until the event loop ends

This post has been edited by Xgoff on Jun 26 2009, 11:59 PM

http://hocuspocus.taloncrossing.com/rii/xlua-dist.zip

Should fix your issue

Also with this release, function registration is redundant. If you tick the 'Auto-Register' box in the properties pane, all your functions will be registered at the start of the frame.

Manual registration is still required if you need to conditionally register a function, register a function later in the application, or require a minimum parameter count.

This release also contains previously mentioned constant values, although they are not yet included in the documentation.

Okay I just added a super-useful feature.

Object.OverlapsBackground() now takes two optional parameters: x offset and y offset. You can test an object as if you had moved it, without actually moving it.

i love you

RII FOR ROOT ADMIN

... this is awesome actually

too bad mmf still waits a frame after a collision before calling collision events... this might still work though since it'll be possible to do a collision check a frame ahead of time

could you do the same thing with the object overlap functions?

maybe for a direct object/object test, but I doubt for the GetOverlapList function because it draws from a list of objects that MMF generated.

that may work out still

also am i doing this incorrectly or

yep i was doing it wrong... lol for some reason i totally skipped over "offset"

This post has been edited by Xgoff on Jun 28 2009, 09:52 PM

a way to get the script and line of a registered function when its called?

elaborate?

On Function "Name"
--> get function execution script
--> get function execution line

would make error reports so much easier

i dunno if it's even possible to get the script since it's compiled

line number maybe

the default lua error reporting seems to be able to so I dunno why the mmf interface shouldn't

edit: for clarification: only the script name and relative path. Not the entire script sdf

This post has been edited by Mecha the Slag on Jul 7 2009, 05:37 PM

it can get basic information about functions and the stack and stuff but afaik that's as far as it goes (that and it will catch syntax errors during compilation, when it still has information about the script)

though i do notice that xlua's error reporting could still use some improvements because both this and lua5.1.exe give a full stack traceback and will tell you WHERE a syntax error is as opposed to just saying "syntax error"

This post has been edited by Xgoff on Jul 7 2009, 05:43 PM

ok i'm probably going to be more annoying with stuff i really don't know about yet but:

how exactly does luagl handle textures? i mean it takes a lua table containing pixel data but i sure as hell hope it converts that to a c array or something and uses that from then on because i don't like the idea of it trying to read hundreds of giant lua tables 60 times per second

it'd almost make more sense if all the functions that require textures could alternatively take a userdata in place of the table... said userdata could be supplied through some function that could load an image via a path (even better if it could have mmf load the image so it'd be possible to use whatever image types mmf supports)

i dunno just a thought