Bug Report Thread

[orange808]

I have a fresh install of bB and Visual Basic on this Win 10 machine (no previous installations to pollute my registry). I followed the instructions. I have the latest versions. I am familar with coding and Windows. The environment variables aren't an alien concept and they are right. Running as admin.

 

So, when I look at the DPC+ examples, they run well enough. But, if I add a set romsize command, they all break. I tried multiple examples. I can change the romsize on Tinkernut all I want.

 

Any suggestions?

[orange808]

Also, will add that this is an Intel x64 machine. Not an AMD or the new ARM x86 Windows emulation. Not an exotic environment.

[+RevEng]

bB DPC+ is only 32k. You can't change the romsize.

[orange808]

​

​

​

bB DPC+ is only 32k. You can't change the romsize.

​

​Fair enough. I found the documentation to be somewhat ambiguous about this. It declares that the romsize is automatically set for you, but I didn't understand that sentence was implying that the romsize could not be changed at all.

​

​How about the multikernel framework with multiple DPC+ 32k parts?

​

​Edit: Apologies. Answered my own question. That won't work. The DPC+ kernel alone is too big to fit inside "containers" of 4k.

​

​

Edited September 15, 2018 by orange808

[+Random Terrain]

Be sure to refresh the bB page.

[bogax]

Not really a bug

 

I am frequently frustrated by having constant macro parameters treated as immediate values.

Not really sure how to deal with that.

How hard would it be to perhaps escape a constant to suppress emission of the "#" ?

(best I could think of)

[+Karl G]

There is a display problem with the rightmost digit of the score that can be exposed with a wide enough score font that I've reproduced on Stella and real hardware. Here's a simple example with the "husky" font with the score = 100. The problem goes away if the player objects are positioned one more pixel to the right (HMP0 = #$D0 and HMP1 = #$E0).

 

 

husky.bas

 

score_graphics.asm

 

husky.bas.bin

[freshbrood]

Is there a way to get rid of the "waterfall effect" when using colored sprites and a colored playfield in the standard kernel?

 

If it were static I wouldn't mind, but it moves and ripples as the player objects move.

[+iesposta]

Request:

This optimization I do that the bB compile passes should.

It is faster assembly and uses less ROM.

 

Can the compile stage find (between anything more complex than “=“) and put all the same value variables together regardless if they are on one line or four, and output more efficient & ROM saving assembly like this?

 

This:

Player0x=0: r=0 s=0 t=0

 

Compiles to this:

lda #0
sta Player0x
sta r
sta s
sta t

 

———

 

And this same thing:
Player0x=0
r=0
s=0
t=0

 

Compiles to this:
lda #0
sta Player0x
lda #0
sta r
lda #0
sta s
lda #0
sta t

 

 

Another example is this one line compiles more efficiently over one value each line:

DF0FRACINC=255: DF1FRACINC=255: DF2FRACINC=255: DF3FRACINC=255: DF4FRACINC=255

[+Karl G]

The multisprite kernel seems to have a problem with the "noscore" option - the timing seems to be off when the score is disabled. I confirmed this with a simple program that just does a drawscreen loop as well as the 9 objects example on RT's site.

 

multi.bas

 

multi.bas.bin

 

 

I have a fix for this issue. The problem happens because a "rts" is skipped when "noscore" is enabled. The original code for noscore in the Multisprite kernel is this:

 ifconst noscore
 jmp skipscore
 endif

I've changed it as follows, which fixes the issue:

 ifconst noscore
 pla
 pla
 jmp skipscore
 endif

Here's a patched version of the kernel for those who need the fix in the meantime:

 

multisprite_kernel.asm

[Sknarp]

My apologies if this has been mentioned already, but I couldn't find anything in the search.
inline ASM throws some warnings- "asm is not a recognized keyword" and more frustratingly "end is a duplicate label definition" if you use more than one block of inline asm, the latter has an annoying tendancy to pop up every time I write any line of code, anywhere in the program, which constantly snaps the window around to focus on where the warning is being generated from.

[+Random Terrain]

40 minutes ago, Sknarp said:

My apologies if this has been mentioned already, but I couldn't find anything in the search.
inline ASM throws some warnings- "asm is not a recognized keyword" and more frustratingly "end is a duplicate label definition" if you use more than one block of inline asm, the latter has an annoying tendancy to pop up every time I write any line of code, anywhere in the program, which constantly snaps the window around to focus on where the warning is being generated from.

 

Are you using the latest version of VbB?

 

https://atariage.com/forums/topic/123849-visual-bb-10-a-new-ide-for-batari-basic/?do=findComment&comment=3537039

 

[Sknarp]

No, apparently I have Build 566, I'll have to update. Thank you RT

[freshbrood]

In Batari Basic, the if !x= statements don't seem to work for me at all. Perhaps I am doing somethhing wrong?

I.e. when I write:

-------------------------------------------

x=2

if !x=1 then COLUBK=Rand


-------------------------------------------

Shouldn't this make the background color change? I could replace "COLUBK" with a goto statement and it still won't go anywhere. "if not" statements seem to be broken/utterly useless.. or am I writing them wrong?

Please help.

[+Random Terrain]

Just now, freshbrood said:

In Batari Basic, the if !x= statements don't seem to work for me at all. Perhaps I am doing somethhing wrong?

I.e. when I write:

-------------------------------------------

x=2

if !x=1 then COLUBK=Rand


-------------------------------------------

Shouldn't this make the background color change? I could replace "COLUBK" with a goto statement and it still won't go anywhere. "if not" statements seem to be broken/utterly useless.. or am I writing them wrong?

Please help.

 

Yep, you're doing it wrong:

 

https://www.randomterrain.com/atari-2600-memories-batari-basic-commands.html#boolean

 

https://www.randomterrain.com/atari-2600-memories-batari-basic-commands.html#if

 

Does "!x=1" work using any programming language? What are you wanting x to be equal to for something to happen?

[freshbrood]

"The not-equal-to operator (!=) is not valid bB syntax. It does not work with batari Basic."


Is there another way to say "if x is not equal to"?


*Edit: My originial post was a typo. I meant to say if x!=


Anyway.. If x<2||x>2 then goto.. seems to work. Same effect. Just uses a bit more memory than I'd like. Oh well.


Thank you RT

Edited September 10, 2019 by freshbrood

[+Random Terrain]

35 minutes ago, freshbrood said:

Anyway.. If x<2 || x>2 then goto.. seems to work. Same effect. Just uses a bit more memory than I'd like. Oh well.

 

If I'm reading that correctly, seems like you'd want to use "If x <> 2 then".

[Mr SQL]

On 6/11/2018 at 1:26 AM, Random Terrain said:

Since I was already in the middle of it, I'm doing my own fix up of your program and I've come across more than one place where you use OR twice in the same if-then. That is not allowed according to the bB page in the Did You Know box. You are allowed only one OR for each if…then statement.

This is an interersting design limitation in the BASIC RT, I'm curious because there is a similar but different design limitation with if's in SuperCharger BASIC -

 

In bB this has to be two statements -

 

 if e=3 or f=8 or g=9 then goto myroutine

 

SuperCharger BASIC allows multiple or's to be in one if statement like that, but you cannot mix or and and in an if statement, I imagine bB would have a similar limitation mixing them if it can only have one or per line.

Edited January 9 by Mr SQL
left out an or

[Sknarp]

The Multisprite Kernel could use a little love, I can't say for sure that it is the Kernel or the emulator the Atari flashback uses but any game using it doesnt work 100% correctly-  I even made a 4k game and with the stock MSK there is still an alignment issue. Everyone including myself thought it was because of my modifications but yeah it happens with a stock MSK.

[+Karl G]

It's definitely the limited emulator in the AFP that causes issues with the Multisprite kernel, and is not a bug in the Multisprite kernel code. The Multisprite kernel uses a technique called an early HMOVE to position the sprites, which hides the black lines on the left side of the screen that you see in many Atari games, but the AFP does not render it correctly. 

[+RevEng]

 

8 hours ago, TwentySixHundred said:

Just wondering if anyone else is have issues setting score fonts after the 1.2 update?

It's throwing me this compile error

segment:  1864          eqm       vs current org: 1894
 free ram: 0
 DPC free RAM= 603

score_graphics.asm.retroputer (3): error: Origin Reverse-indexed.

Errors were encountered during assembly.

 

 

Thanks for the report @TwentySixHundred. Looks like I introduced a bug with the hex font addition. It's fixed in the github master. I've attached the fixed score_graphics.asm here, in case anybody needs it.

 

I'll do an official github release in a week or two, to allow other issues to be reported and fixed.

 

score_graphics.asm

[TwentySixHundred]

15 minutes ago, RevEng said:

 

 

Thanks for the report @TwentySixHundred. Looks like I introduced a bug with the hex font addition. It's fixed in the github master. I've attached the fixed score_graphics.asm here, in case anybody needs it.

 

I'll do an official github release in a week or two, to allow other issues to be reported and fixed.

 

score_graphics.asm 4.24 kB · 2 downloads

Thanks for the prompt fix 👍

[TwentySixHundred]

Hi guys it was brought to my attention by @sramirez2008 and @Karl G that my game is crashing with this error message.

After having a quick search on AA i have noticed this thread but there was no solution as to what actually was the issue. Does anyone know Is it was bB DPC+ bug or Stella or coder error? The crash also happened real hardware using the harmony cart so im assuming it's nothing on the Stella side of things. Before i left the house i enabled the Dev settings and sure enough it did the same.

 

[+RevEng]

My best guess without source, is that you're indirectly feeding the DPC÷ ARM bad data. Have you stepped on DPC memory,, directly written to arm queues, or otherwise done any under the hood access?

[TwentySixHundred]

9 minutes ago, RevEng said:

My best guess without source, is that you're indirectly feeding the DPC÷ ARM bad data. Have you stepped on DPC memory,, directly written to arm queues, or otherwise done any under the hood access?

Nothing that i know of other then collision detection with playfield and my fake "collision detection" when the warheads hit the ground. This led me to think the sub routine after was causing it. Sure enough i comment out a line and it works as "normal" however it will then crash at random without even executing the troublesome code. It's a strange one that's for sure, i have sent you a DM