Symmetry of TNG

Short about some uart related problems i know of: Spock have been sitting quite some time with doom tryig to get the "$80 before packet" thing to work... but i think he gave up on that.. since something that simple did not work properly with the doom code... even though it should have worked! other parts of the doom code then became unstable and crashed.. ????!!!! (insert a steel bar into a card-house and expect it to be more robust?).. So I think he agreed to start from scratch with the net code, not try to build it via doom.. (or that is what i would suggest to do..). Personally i have 2 jaguars communicating via rs232 using the uart IRQ!!... BUT!... as soon as there is an error in the transmission the uart IRQ hangs!... and it is impossible to recover it, except via a reset/poweroff... If there are 1object on screen it works nice... 2 or 3 then the hang occures more often... totally unreliable... I only use the 68K irq... perhapps i need to go into a gpu&dsp combination.. (my plan al long anyhow). One thing i know though, is that the programmes that use the uart and get something out of it, use "manual polling" and not the irq. ....(cf Doom, aircars... ...not BS, they use either gpu or dsp... that im sure of). Either there are some stupid problems in my IRQ handler (especially when an error occures) or the irq is unstable.... (or i would be able to clear error & reset it right?)... Its tiresom to debug this sh-t since it takes two trail-n-error uploads each time... And it was almost a year since I did any work on it... and it take allot of time, time i would rather "waste" on something else... (polygon). Anyone else have any experience?

Well as long as the graphisist dows not create any single horisontal white line (odd frame white, even frame black) it should be ok... but usual graphis usually never contain that. Hmm. perhapps my code wasnt as versetile as i thought.. though the trickery I do to setup the registers probably need to change for 60Hz.. thats true.. regarding "OP interlace bug": there are a number of ways to make this: --set dwidth to twice the size of the picture, but only display width pixels... then each vbl change the start position of the first pixel. (the one i used) --The 2nd a bit more "hacker style" woudl be to do a pre-processing of the picture separating the odd&even lines into two pictures "on top" of another.. So 1st picture (say 640 width.. and also dwidth) only contains even lines... the secound picture (640*480*2 bytes away in memory) contains the odd lines... THEN! ..you could actually utilise the other OP bug in your favour and you would only have to update the height variable in the OPL every other vbl! =) Hence saving OPL update time!... Thats the hacker way to do interlace >So you disassembled my code, uh ? *caugh* ..I have an alpine.. it comes as default But since Im the 1st to do interlace on the jaguar (as far as i know) I am the one that has to check where the code came from to see if there was a leak ...but it checked out ok Eor=quick.. well i use the vbl counter and test the lowest bit.... "same same but different" =) cheers! -------------------------------- to TXG: >Cool so 640 x 480 with borders could become a new standard then Borders? ....Only if you whant to do a OldSchool ST tribute demo on the jaguar ...otherwise No bordes requiered!... I would say a picture 720*576 would absolutely fill the hole screen... and it would be magnificantly nice! =) Just One problem... and i guess thats why atari never realy considered it... 720*576*2=829440 bytes!!!.. its like half the avaiable memory just for one screen!... (double buffer and you have 4K of gpu mem for your game.. Yeeehh! For a demo this is Gold!... perhapps for a menu...anywhere wer you can sattle for 1screen .....but for a double buffer game? ....nahh... but it is nice to know the jaguar can do what every other system can do.. interlace... now we just have to ues it in a production! =) cheers! /Sym

..could anyone give a short english recapitulation of what you people have come up with so far? ...Im just qurious... And preferable if anyone tested it in real life with code (and not just theory.... because in theory I have a good understanding of what causes the error.. but my experience shows that theory & practise dont realy mix). Thanx! /Sym

Thanx for the email zerosquare.. ..hmm my routine had the same kind of "flicker" in the text.. I guess it is because the text is so thin. Guess interlace gfx needs to be made with some restrictions...to not flicker that much. Some thoughts: --Hmm i never tested my rout in 60Hz, but it also didnt work.. ?.. was a bit surpriced.. since the method has nothing to do with the framerate, so i kind of assumed it would work.. but did not.. So if someone wants to make a "European game ONLY" then here is a way to do it =) It needs to be debugged in 60Hz to se if it is possible to do Highres on an US jaguar... --It is said that the OP in interlace mode skipps every other line.... but that i never got to work, so i worked my way around it.. what about you? --and then there is that "interlace vbl irq" thing... but that i guess you figured out ... (i dont understand that eor you do but aparently it works =) regards /Sym

=) Well i mean IF you can display a full gfx picture say 640x480 in nice interlace just by setting that register, then NO! there is no more work... (then i must have done something wrong, and wasted time finding a workaround for my own mistake).. I guess Im, "between the lines" asking: can you display such a picture? =) just by setting that register?

hmm... well if it were that simple I would not have had to waste some week on debugging it (when I was doing this (december -04). And it is not as easy as just setting a register (like it is on the falcon).. atleast it was not for me... And there are several ways to get "interlace" ..on the jaguar but only one of them will produce "True" interlace... that is an actual improvement in display quality due to the fact that the display is shifted half a scanline up/down on the TV. You can se this if you write a interactive program that changes these things on user commands, and then sit with a lupe infront of the TV and stare at the RGB dots, press and se that they actually shift... 8) ...phuu it was no easy task for me atleast.... (this should actually be a separate topic, so i stop =)

..... ok i will not tell you anything more than.. according to the ataris docs we have there is no way to turn it on... No i will stop spoiling your fun back to whistling...

Hmm... interesting... question... I always wonderedhow accurate the schematics are... but we can never be sure since its just to much work to trace all of the lines.. anyhow.. sure noone can se difference between 8 & 15milj colors.. BUT... It would defenitely be possible to se difference if ONLY blue was used!... say draw a screen size shaded line, from black to full blue... then the "increments" should defenitely be seen with 7bits (127 shades) ...and it should be smoot on a full 8bit blue... (I remember this from playing with APEX color modes for the falcon... teh TC mode was never as good as their interlace "24bit mode" ..(interlaced 256colors). Similar should be possible to determine on the jaguar... (if now anyone cares =) ....well.. ok... now with the CF 720*480 24bit images would be possible... (since they eat up to much memory otherwise Hmm... speaking of i-lace.... ...i have an unreleased secret ;P ..waiting for the right moment... cheers /Sym

Last time i tested dosbox was when it came up howto use wdb & an alpine dev env under xp... but then dosbox didnt have PPport support.....? so it was pritty pointless to use it for alpine dev. .. But i have a pure W98 mashine (LCD PC) that works now.. even though I would like to get it running under xp.. and the only way i know howto is via a w98 emulator like wmware... ahh well.. sorry i couldnt help more. cheers

Hi Ray! DosBox? ....ahhmm... unless they did some quantum leap coding to get it out of the nonworking beta stage I would sugest WMWare instead... (viritual pc perhapps) .....I know Mr.Spock uses WMware and runs w98 and it works great..(even with an alpine). (iirc wmware is now freeware (?). anyhow I hope you use the latest environment, if not you can get it here: http://www.cs.umu.se/~mat97jkn/tng/ and if someone have newer versions of the programms, please concact either one of us.. Perhapps that setup can help you ? ......(would also like an independent person to try and install it and se if it works (i dont remember if he wrote a tutorial on how to install it or not...). apparently all ataris C programs compile ok in the code (he did alot of work to get them to compile) ...and iirc even the gpu C compiler works... hope it helps. cheers /Sym

And the devpac result of the exact same code is 46Kb? ....verry odd... you must have some flag set somewhere to include... stuff for the debugger or something... how are your mac/aln lines defined? took some from an old code and my lines that work on bjl look something like this: MACFLAGS = -v -fb -g -rd ALNFLAGS = -n -v -v -v -rd -o bootcd.prg -a $(STADDR) x x I have never compared the result with devpac.. but my point beeing that when you code assembler the compiler should produce the exact same instruction as you wrote no matter what compiler used... (..unless perhapps devpac has some optimisation flags set.... but not raw code!.... Ahh.. wait... it might be that they differ in the data definitions... if you have a ".bss" are on devpac that part is never included in the compile... if you dont the dataspace will be included in file size making it huge.. perhapps you have something similar on madmac... but on devpac you have the .bss statement.. ? ahh well... will not disturb your connexion coding! thats more important =) cheers

What? ....are you saying that ALN created ~3times as big code? ...(compiling the exact same code as on devpac)?.... If so then.... i would realy like to know what fills the other two 46Kb parts to make it 146Kb... This is interesting but horrifying... that are just not supposed to happen... not when compiling assembler... (if it was C i could understand).. If you can check out the difference between the 2 codes and see what differs that would be nice to know. regards

The scans i mentioned. OBS for personal, private use only! ...as a precaution its a pass´d zip. articles are rather old, but a good introduction to simpler form of AI. some keywords: -deterministic chase algo. -random logic -Finite State Mashines -Environmental Sensing -Memory & learning These are just quick intros into the topics so you can dig deeper once you learn what you are looking for in your specific game. Includes some simple C sources. /Sym AI_GameDev.zip

>And why not using I.A. in a shoot game ? I did not say that one Soulg follow classical mechanics ...I would defenitely do AI in a shotemup. But one that does not require to much computing power.. There are some simple things that can be used to simulate "intelligence" and that is randomly choose a behaviour. ie in a shotemup.. have a list of predefined paths for the aliens to follow. IF the aliens die to fast, randomly choose another path for them to follow. it will "appear" to be based on a choise that they understand they are dying to fast and need to change strategy to survive. And it costs almost nothing in cpu power... just a rand% and some compares... This is the simplest form of AI and for allot of "predefined path" things it works great... ... UNLESS you are doign AvP2 ...=) remember "Bugs" are not intelligent (no matter what alien3,4,5,6,7,... may try to fool you ...they go straight at you, every alien for himself, untill last one.. ant that is how it should be (..no this is a joke, and i dont want to turn it into a avp related topic so dont start the flame thrower I will c if i can scan some pages out of a magazine and post them later, about AI... that explains the basics and in theory how to go beyond... (with some c source iirc) cheers

I think it depends on what type of game you are thinking of doing.. I mean if it is a sprite game, shotemup, then the classical "lets follow a nice sine curve.. I mean STICK TO FORMATION no matter what!" =) ...is the usual thing to do. the "advanced" stuff then usually go for "the chase" algo. ie move the foe in the direction of the vector between player & foe.... (lets rename foe to alien, sounds much better =) But thats is no A.I either... A avp thingy would need some kind of "deterministic" algorithm (if i remember correctly(?..probably not). It needs to know of its environment, have properties like "good or bad" hearing to react if player makes sounds... I mean it all comes down to allot of "IF player then Alien" segments.. Adding a random variable of what alien actually does in cases makes "intelligence" apperence. Besides that there is offcourse the more advanced topics (the ones I read about). Fuzzy Logic, for strategic games is verry interesting... would work for any "desision making" type of game... (ex. when alien should flee for his life due to the fact that he is beeing slauterd =) and stuff like that. Even higher up: Neural/nervous networks.. only way to true A.I ...alien will learn how you play and adapt, in the end get unbetable ...depending on size of network offcourse... This is also related to the problem of finding a path between to points... (that im sure the robot builder here knows allot about ...and a path that doesnt get stuck... (seeing the scene as a "vector" field, or a hight map, and all things that are inpenetrable is either high or has vectors pointing away from them.. Ahh.. enough lunchtalk.. =) ..the short question would have been "What kind of game are you thinking of?"... =) cheers /Sym

Finally! =) I want to thant the following people for helping me with this resigration: babel.altavista.com hehe =) ....without its totally wierd translations I would not have made it here also GT for convincing me to join ..no.. just thought to say Hi to all.. most people in this english department already know me.. The frensh people know my jaguar demos from connexion (was it 2003?) ..but they seem to have forgotten me... ahh well.. Hopefully I will be active posting jaguar coding realted "junk" so people here will learn who i am I will just say that I will keep to mostly posting coding related posts.. since jaguar "smalltalk" i will keep to AtariAge... (I want this to be coding related and not just another smalltalk forum, correct me if i am wrong). ahh well... its Hacker weekend and i need some source to get into the moode Merci tout l'espoir que ce sera un jurney prospère ( i could just as well say "blablabla to you all" since i have no idea what it says /Symmetry ...now stumbling on frensh forums.