Forums

I've tried a few times to get up and running with Sega Master System programming. I followed the setup for with ConTEXT, poured over the code, compiled the program, wrote it to an SD card, fired it up on my hardware etc.. all was going well until the tutorial abruptly stops. These tutorials seemed like they could have been an absolutely invaluable resource and I wonder if there is any plan to continue them?

Regards

I guess you mean at https://www.smspower.org/maxim/HowToProgram - the answer is, probably one day but maybe not very soon.

I hope so too, It helped a lot on setup.

This one actually helped me get going on a full game, though I wish we had some good tutorials on NPC setup and behavior scripting.

If you want to continue try this, its just as helpful.
https://www.smspower.org/Articles/CreateARacingGame

I’m doing all my dev in VSCode and that’s x-platform and way nicer than ConText + batch files. So yeah, it could use an update.

@Maxim - I know these things are a lot of effort but really, I think that this type of effort is really awesome for the historic preservation and I would love to be able to get involved in this stuff. Hope to see some tutorials in the future. :)

@IllusionOfMana - That link is incredible so thank you!!

@SavagePencil - I'm usually a TypeScript dev these days and I work with VSCode on a daily basis. Having a similar dev environment would be excellent. Do you have any instructions for getting setup for SMS dev on VSCode?

SavagePencil wrote

I’m doing all my dev in VSCode and that’s x-platform and way nicer than ConText + batch files. So yeah, it could use an update.

Do you have any tutorial or can overview how you have this set up? I'd be most interested to have a similar setup :)

Blaghhh yeh I’ll post something tomorrow.

You can get a sneak peek at the environment here:

https://github.com/SavagePencil/SMSSampleApp?files=1

Personally I’ve dropped ConTEXT and moved to Notepad++. I’ve found the “modern” editors too slow for my common use case of 10MB+ disassemblies, but haven’t tried VSCode specifically.

I like the flexibility of project management, git integration, and being able to launch the assembler and emulator all from the IDE. I've written up my notes on how to do that here.

I actually found ConTEXT easy to setup, though Notepad++ is too. I always struggle to set up other dev environments like SGDK. Though there is a nice web based IDE at 8bitworkshop that looks promising and easy to use. IDK how far you can go with it just yet though.

I use Geany (C development) or Eclipse (asm development) with efry's z80editor plugin.

Geany is quite good. Approach is quite similar to ConTEXT.
Eclipse... well, I hate that, but efry plugin is the reason why I'm using it.

I headed over to the SMSDevKit, as I thought perhaps in the mean-time I'd look at programming in C. Got a great response there and I thought that response might be useful here: https://github.com/sverx/devkitSMS/issues/12

I have since setup a C environment in Visual Studio Code and tonight I will attempt to compile the sources from: https://www.smspower.org/Homebrew/AstroForce-SMS

Many thanks for the comments.

82223782_10156887448261149_2830841411836837888_n.jpg (30.44 KB)

In particular, one tutorial I'd like to see is a worked example of how to design an actor struct such that its fields can be stepped through sequentially with HL. This way I wouldn't have to do random access through slow (IX+5) all the time. It seems all the 8080-family programming tutorials that I've seen stop before they get to this point.

well you could have actors variables at 256-byte aligned address, so you would do

ld h,>actor
ld l,member
ld a,(hl)

instead of

ld ix,actor
ld a,(ix+member)

and to access next member, if you organized them in convenient way, you just increment L instead of HL

there are caveats of course, but that's surely quicker

...and if you have your actor members in a struct, you can easily do this (even if it may not be optimal):

.STRUCT sActor
XPos DB
YPos DB
HitPoints DW
.ENDST

; Put this in a RAM section to pre-alloc 4 bytes of the sActor struct.
.DSTRUCT MyActor INSTANCEOF sActor

ld hl, MyActor
.REPT sActor.HitPoints
inc hl
.ENDR

(you could get fancy and find the break-evens for when it's better to add hl, de vs inc hl)

One of my current NES projects has 21 bytes per actor of striped state tables, which would be equivalent to the following struct:

.STRUCT sActor
XVel DW ; in pixels per 256 frames
XSubpx DB ; subpixel component of position
XPos DW ; pixel and "screen" component of position
YVel DW
YSubpx DB
YPos DW
ClassID DB ; determines which move routine to call
AnimFrame DB
FrameTimer DB
StunTimer DB ; time until hitstun expires
FaceDir DB
HitPoints DB
DamageHt DB ; height at which it most recently took damage
CollideHt DB ; height of hurtbox
CollideBits DB ; some states temporarily disable hit/hurtbox
Strength DB ; number of hit points removed from player on collision
VramBase DB
.ENDST

For a struct this big, I imagine access would fall fairly regularly on the add hl, de side of the break-even point unless the fields can somehow be optimally ordered. How would one go about determining this optimal ordering in advance?

A memory layout following the paradigm suggested by sverx would use 21 bytes of each 256-byte page for the actor struct and (256 - 21) = 235 bytes for something else. What support is there in the assembler or in any other tool for divvying up that something else?

Some of my actors are bigger than that.
Try to round it up (to 8,16, 32, 64), align datas in ram for a faster access or modifications.

Sometimes it's better to use (ix+n) that add hl, de.

You can try to place posy, posx and meta (in this order) as they are needed to be pushed in the SAT.

PinoBatch wrote

A memory layout following the paradigm suggested by sverx would use 21 bytes of each 256-byte page for the actor struct and (256 - 21) = 235 bytes for something else. What support is there in the assembler or in any other tool for divvying up that something else?

in WLA-DX you can use 256-byte aligned free ramsections for the former and regular (unaligned) free ramsections for the latter

PinoBatch wrote

For a struct this big, I imagine access would fall fairly regularly on the add hl, de side of the break-even point unless the fields can somehow be optimally ordered. How would one go about determining this optimal ordering in advance?

I haven't tried it myself, but I bet you could do something along the lines of:

.IF sActor.myMember > 5 ;Or whatever the break-even point between inc and add hl, de is
ld de, sActor.myMember
add hl, de
.ELSE
.REPT sActor.myMember
inc hl
.ENDR
.ENDIF

Not sure if you could .MACRO this.

Author	Message
Blitmaster Joined: 23 Aug 2005 Posts: 37 Location: Manchester, UK	Will there ever be more programming tutorials? Posted: Sun Dec 22, 2019 8:08 pm
	I've tried a few times to get up and running with Sega Master System programming. I followed the setup for with ConTEXT, poured over the code, compiled the program, wrote it to an SD card, fired it up on my hardware etc.. all was going well until the tutorial abruptly stops. These tutorials seemed like they could have been an absolutely invaluable resource and I wonder if there is any plan to continue them? Regards

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14726 Location: London	Posted: Sun Dec 22, 2019 9:50 pm
	I guess you mean at https://www.smspower.org/maxim/HowToProgram - the answer is, probably one day but maybe not very soon.

IllusionOfMana Joined: 08 Sep 2018 Posts: 270	Posted: Sun Dec 22, 2019 9:54 pm
IllusionOfMana Joined: 08 Sep 2018 Posts: 270	I hope so too, It helped a lot on setup. This one actually helped me get going on a full game, though I wish we had some good tutorials on NPC setup and behavior scripting. If you want to continue try this, its just as helpful. https://www.smspower.org/Articles/CreateARacingGame

SavagePencil Joined: 23 Aug 2009 Posts: 213 Location: Seattle, WA	Posted: Mon Dec 23, 2019 12:52 am
	I’m doing all my dev in VSCode and that’s x-platform and way nicer than ConText + batch files. So yeah, it could use an update.

Blitmaster Joined: 23 Aug 2005 Posts: 37 Location: Manchester, UK	Posted: Mon Dec 23, 2019 1:01 am
	@Maxim - I know these things are a lot of effort but really, I think that this type of effort is really awesome for the historic preservation and I would love to be able to get involved in this stuff. Hope to see some tutorials in the future. :) @IllusionOfMana - That link is incredible so thank you!! @SavagePencil - I'm usually a TypeScript dev these days and I work with VSCode on a daily basis. Having a similar dev environment would be excellent. Do you have any instructions for getting setup for SMS dev on VSCode?

Blitmaster Joined: 23 Aug 2005 Posts: 37 Location: Manchester, UK	Instructions? Posted: Fri Dec 27, 2019 5:55 pm
	SavagePencil wrote I’m doing all my dev in VSCode and that’s x-platform and way nicer than ConText + batch files. So yeah, it could use an update. Do you have any tutorial or can overview how you have this set up? I'd be most interested to have a similar setup :)

SavagePencil Joined: 23 Aug 2009 Posts: 213 Location: Seattle, WA	Posted: Sat Dec 28, 2019 1:11 am
	Blaghhh yeh I’ll post something tomorrow. You can get a sneak peek at the environment here: https://github.com/SavagePencil/SMSSampleApp?files=1

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14726 Location: London	Posted: Sat Dec 28, 2019 8:00 am
	Personally I’ve dropped ConTEXT and moved to Notepad++. I’ve found the “modern” editors too slow for my common use case of 10MB+ disassemblies, but haven’t tried VSCode specifically.

SavagePencil Joined: 23 Aug 2009 Posts: 213 Location: Seattle, WA	Posted: Sat Dec 28, 2019 6:23 pm
	I like the flexibility of project management, git integration, and being able to launch the assembler and emulator all from the IDE. I've written up my notes on how to do that here.

IllusionOfMana Joined: 08 Sep 2018 Posts: 270	Posted: Sat Dec 28, 2019 8:09 pm
IllusionOfMana Joined: 08 Sep 2018 Posts: 270	I actually found ConTEXT easy to setup, though Notepad++ is too. I always struggle to set up other dev environments like SGDK. Though there is a nice web based IDE at 8bitworkshop that looks promising and easy to use. IDK how far you can go with it just yet though.

sverx Joined: 05 Sep 2013 Posts: 3794 Location: Stockholm, Sweden	Posted: Sat Dec 28, 2019 8:47 pm
	I use Geany (C development) or Eclipse (asm development) with efry's z80editor plugin. Geany is quite good. Approach is quite similar to ConTEXT. Eclipse... well, I hate that, but efry plugin is the reason why I'm using it.

Blitmaster Joined: 23 Aug 2005 Posts: 37 Location: Manchester, UK	Visual Studio Code Posted: Mon Jan 13, 2020 4:49 pm
	I headed over to the SMSDevKit, as I thought perhaps in the mean-time I'd look at programming in C. Got a great response there and I thought that response might be useful here: https://github.com/sverx/devkitSMS/issues/12 I have since setup a C environment in Visual Studio Code and tonight I will attempt to compile the sources from: https://www.smspower.org/Homebrew/AstroForce-SMS Many thanks for the comments. 82223782_10156887448261149_2830841411836837888_n.jpg (30.44 KB)

PinoBatch Joined: 05 Dec 2019 Posts: 56 Location: USA	Posted: Wed Feb 05, 2020 2:15 am
PinoBatch Joined: 05 Dec 2019 Posts: 56 Location: USA	In particular, one tutorial I'd like to see is a worked example of how to design an actor struct such that its fields can be stepped through sequentially with HL. This way I wouldn't have to do random access through slow (IX+5) all the time. It seems all the 8080-family programming tutorials that I've seen stop before they get to this point.

sverx Joined: 05 Sep 2013 Posts: 3794 Location: Stockholm, Sweden	Posted: Thu Feb 13, 2020 3:43 pm
	well you could have actors variables at 256-byte aligned address, so you would do ld h,>actor ld l,member ld a,(hl) instead of ld ix,actor ld a,(ix+member) and to access next member, if you organized them in convenient way, you just increment L instead of HL there are caveats of course, but that's surely quicker

SavagePencil Joined: 23 Aug 2009 Posts: 213 Location: Seattle, WA	Posted: Thu Feb 13, 2020 4:00 pm
	...and if you have your actor members in a struct, you can easily do this (even if it may not be optimal): .STRUCT sActor XPos DB YPos DB HitPoints DW .ENDST ; Put this in a RAM section to pre-alloc 4 bytes of the sActor struct. .DSTRUCT MyActor INSTANCEOF sActor ld hl, MyActor .REPT sActor.HitPoints inc hl .ENDR (you could get fancy and find the break-evens for when it's better to add hl, de vs inc hl)

PinoBatch Joined: 05 Dec 2019 Posts: 56 Location: USA	Posted: Wed Feb 19, 2020 4:20 am
PinoBatch Joined: 05 Dec 2019 Posts: 56 Location: USA	One of my current NES projects has 21 bytes per actor of striped state tables, which would be equivalent to the following struct: .STRUCT sActor XVel DW ; in pixels per 256 frames XSubpx DB ; subpixel component of position XPos DW ; pixel and "screen" component of position YVel DW YSubpx DB YPos DW ClassID DB ; determines which move routine to call AnimFrame DB FrameTimer DB StunTimer DB ; time until hitstun expires FaceDir DB HitPoints DB DamageHt DB ; height at which it most recently took damage CollideHt DB ; height of hurtbox CollideBits DB ; some states temporarily disable hit/hurtbox Strength DB ; number of hit points removed from player on collision VramBase DB .ENDST For a struct this big, I imagine access would fall fairly regularly on the add hl, de side of the break-even point unless the fields can somehow be optimally ordered. How would one go about determining this optimal ordering in advance? A memory layout following the paradigm suggested by sverx would use 21 bytes of each 256-byte page for the actor struct and (256 - 21) = 235 bytes for something else. What support is there in the assembler or in any other tool for divvying up that something else?

ichigobankai Joined: 04 Jul 2010 Posts: 540 Location: Angers, France	Posted: Wed Feb 19, 2020 6:53 am
	Some of my actors are bigger than that. Try to round it up (to 8,16, 32, 64), align datas in ram for a faster access or modifications. Sometimes it's better to use (ix+n) that add hl, de. You can try to place posy, posx and meta (in this order) as they are needed to be pushed in the SAT.

sverx Joined: 05 Sep 2013 Posts: 3794 Location: Stockholm, Sweden	Posted: Wed Feb 19, 2020 2:03 pm
	PinoBatch wrote A memory layout following the paradigm suggested by sverx would use 21 bytes of each 256-byte page for the actor struct and (256 - 21) = 235 bytes for something else. What support is there in the assembler or in any other tool for divvying up that something else? in WLA-DX you can use 256-byte aligned free ramsections for the former and regular (unaligned) free ramsections for the latter

SavagePencil Joined: 23 Aug 2009 Posts: 213 Location: Seattle, WA	Posted: Wed Feb 19, 2020 4:31 pm
	PinoBatch wrote For a struct this big, I imagine access would fall fairly regularly on the add hl, de side of the break-even point unless the fields can somehow be optimally ordered. How would one go about determining this optimal ordering in advance? I haven't tried it myself, but I bet you could do something along the lines of: .IF sActor.myMember > 5 ;Or whatever the break-even point between inc and add hl, de is ld de, sActor.myMember add hl, de .ELSE .REPT sActor.myMember inc hl .ENDR .ENDIF Not sure if you could .MACRO this.

Forums

View topic - Will there ever be more programming tutorials?