TheBackShed.com

Forum Index : Microcontroller and PC projects : CMM2: V5.05.03 Bug Reports

Page 1 of 4

Author

Message

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 8579

Posted: 10:52am 22 Jun 2020

Copy link to clipboard

Print this post

Please report any bugs in the release version of 5.05.03 here.

Current issues:

Quote If you draw a box or a circle that is entirely off the screen, the edge still appears on screen.

TBC:

Quote I believe I spotted a CMM2 Basic bug. SPRITE LOAD �spritefile.spr� appears to mangle sprites in any color mode other than 8 bits.

This bug or something like it was recently fixed - awaiting more info before investigating

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 11:45am 22 Jun 2020

Copy link to clipboard

Print this post

Sorry Peter, but is the issue with a program "RUNning" itself resolved? I didn't see anything in the 5.05.03 release notes that seemed to correspond.

Thanks for resolving the other issues I reported.

Regards,

Tom

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 8579

Posted: 12:03pm 22 Jun 2020

Copy link to clipboard

Print this post

Quote is the issue with a program "RUNning" itself resolved?

It is not something I can look at unless someone can demonstrate it in a simple enough program to investigate. I read your release notes on zmim to mean it was no longer an issue.

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 12:09pm 22 Jun 2020

Copy link to clipboard

Print this post

matherp said It is not something I can look at unless someone can demonstrate it in a simple enough program to investigate.

You obviously missed it in the kerfuffle over SD cards, it's on p15 of the previous thread:

Function exec(a) As Float
�Local s$
�Do
� �Line Input "Run again [y|n] ", s$
� �s$ = LCase$(s$)
�Loop Until s$ = "y" Or s$ = "n"
�If s$ = "y" Then Run
�exec = 1
End Function

Sub main()
�Cls
�a = exec(1)
End Sub

main()

Run and answer "y" followed by "n" to get this error:

[17] Error: Nothing to return to

It's not the same error message, but it has the same "feel" of the interpreter having lost context so I suspect it's the same issue.

Quote I read your release notes on zmim to mean it was no longer an issue.

Yes, I worked around it, but I thought it might still be an issue that should be resolved for the future.

Regards,

Tom
Edited 2020-06-22 22:10 by thwill

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 8579

Posted: 12:18pm 22 Jun 2020

Copy link to clipboard

Print this post

Yes - missed it.

Just checked and it is in the MM2 as well so I'll pass it over to Geoff

UPDATE

The problem is fixed by putting an END statement after main(). Please try the same on zmim V1. I don't have a copy.
Edited 2020-06-22 23:54 by matherp

Geoffg

Guru

Joined: 06/06/2011
Location: Australia
Posts: 3165

Posted: 01:59pm 22 Jun 2020

Copy link to clipboard

Print this post

the issue with a program "RUNning" itself

What a contrived case!

I can guess as to what is causing this and will add it to the todo list. In the meantime don't run a program from within a sub.

Geoff

Geoff Graham - http://geoffg.net

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 8579

Posted: 05:53pm 22 Jun 2020

Copy link to clipboard

Print this post

New bug

GPS function not working at the command line but will work in a program

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 08:15pm 22 Jun 2020

Copy link to clipboard

Print this post

matherp said The problem is fixed by putting an END statement after main(). Please try the same on zmim V1. I don't have a copy.

That doesn't fix it for the real code. My guess is that in the "contrived example" whatever bad state the interpreter/firmware ends up in has by luck come across that END statement which allows it to end "gracefully", whereas in the real code it hits something that it can't deal with, i.e.

[execute.inc:24] Error: EXEC is not declared

where it looks like it is trying to assign to the EXEC function which it no longer knows about in its bad state.

Perhaps it also matters how deep the call stack is when END is called.

Regards,

Tom

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 08:16pm 22 Jun 2020

Copy link to clipboard

Print this post

Geoffg said What a contrived case!

Damned if I do and damned if I don't

. The real case isn't contrived but it is hundreds of lines long.

Best wishes,

Tom
Edited 2020-06-23 06:17 by thwill

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 08:40pm 22 Jun 2020

Copy link to clipboard

Print this post

thwill said That doesn't fix it for the real code.

Ignore me I'm talking twaddle.

Ensuring that RUNned program actually invokes the END command instead of just "running off the end of the code" does fix the problem.

Thanks,

Tom
Edited 2020-06-23 06:58 by thwill

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 12:50pm 23 Jun 2020

Copy link to clipboard

Print this post

Possible BUG? conditionals don't perform short-circuit evaluation

Try this:

Function always_true()
�Print "always true"
�always_true = 1
End Function

Function always_false()
�Print "always_false"
�always_false = 0
End Function

Cls
Print always_true() Or always_false()
Print always_false() And always_true()

The output is:

always true
always_false
1
always_false
always true
0

I would have expected:

always true
1
always_false
0

My expectation comes from the fact that if the first operand of an OR is true, or the first operand of an AND is false then the code does not need to evaluate the second operand.

As it stands you can't do this:

If tk_char = -1 Or Chr$(tk_char) = " " Then ...

Because Chr$() will report an error when tk_char = -1

I know there are ways to workaround it, but (in a more complex case than this) they are more verbose.

Best regards,

Tom
Edited 2020-06-23 22:51 by thwill

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 8579

Posted: 01:03pm 23 Jun 2020

Copy link to clipboard

Print this post

Everything to do with the core MMBasic language is for Geoff and not me and will probably be common across all V5 MM and possibly V4.5

For this sort of issue can I request that you also test on a MM2 and only post on this thread if you see a difference in behaviour (meaning I've done something stupid). If the behaviour is the same it is probably better to raise the issue on a separate thread for Geoff to comment.

I will keep the CMM2 up-to-date with any of Geoff's changes to core MMBasic but won't be making any changes that cause behaviour to deviate from other versions.

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 01:37pm 23 Jun 2020

Copy link to clipboard

Print this post

matherp said Everything to do with the core MMBasic language is for Geoff and not me and will probably be common across all V5 MM and possibly V4.5

OK, but Geoff isn't so foolhardy as to have a BUG reporting thread

Quote For this sort of issue can I request that you also test on a MM2 ...

It would be my pleasure if you'd like to send me one

in the meantime perhaps someone else will oblige.

Quote ... and only post on this thread if you see a difference in behaviour (meaning I've done something stupid). If the behaviour is the same it is probably better to raise the issue on a separate thread for Geoff to comment.

I will keep the CMM2 up-to-date with any of Geoff's changes to core MMBasic but won't be making any changes that cause behaviour to deviate from other versions.

Ack.

Best wishes,

Tom

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

Turbo46

Guru

Joined: 24/12/2017
Location: Australia
Posts: 1593

Posted: 02:10pm 23 Jun 2020

Copy link to clipboard

Print this post

matherp said but won't be making any changes that cause behaviour to deviate from other versions.

The FILES command is an exception?

Bill

Keep safe. Live long and prosper.

CircuitGizmos

Guru

Joined: 08/09/2011
Location: United States
Posts: 1421

Posted: 02:56pm 23 Jun 2020

Copy link to clipboard

Print this post

The output is:
always true
always_false
1
always_false
always true
0

This is what I expect. Given that the functions need to run in order to return either a 0 or a 1, anything inside the function has to be executed.

Micromites and Maximites! - Beginning Maximite

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 03:07pm 23 Jun 2020

Copy link to clipboard

Print this post

CircuitGizmos said This is what I expect. Given that the functions need to run in order to return either a 0 or a 1, anything inside the function has to be executed.

Thanks for confirming. That's not actually the case for the programming languages I'm familiar with which would not evaluate the second operand when its value cannot change the result.

However looking at VB.NET in the table at Short-circuit evaluation it looks like this is not a bug for BASIC, but we are "missing" the AndAlso and OrElse operators.

Regards,

Tom
Edited 2020-06-24 01:08 by thwill

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 8579

Posted: 03:34pm 23 Jun 2020

Copy link to clipboard

Print this post

Quote The FILES command is an exception?

Files in not part of "core" MMBasic - it doesn't even exist on a MM2m, I'm talking about the main language elements. In general commands that can only be run at the command line are not "core"

CircuitGizmos

Guru

Joined: 08/09/2011
Location: United States
Posts: 1421

Posted: 11:58pm 23 Jun 2020

Copy link to clipboard

Print this post

thwill said
Thanks for confirming. That's not actually the case for the programming languages I'm familiar with which would not evaluate the second operand when its value cannot change the result.

If you are talking a compiled language, then that would be optimized out. MMBasic is interpreted and cannot act another way.

Micromites and Maximites! - Beginning Maximite

thwill

Guru

Joined: 16/09/2019
Location: United Kingdom
Posts: 3841

Posted: 12:06am 24 Jun 2020

Copy link to clipboard

Print this post

CircuitGizmos said

If you are talking a compiled language, then that would be optimized out. MMBasic is interpreted and cannot act another way.

I don't think you are correct, consider:

a = fun1() AND fun2()

When the interpreter reads fun1() it evaluates it and stores the value, it then reads AND, if the stored value is 0 then it would be permissable for the interpreter to read but not evaluate the second operand since it knows the result of the evaluation cannot change the result of the expression.

Case in point Python only has short-circuit boolean operators yet it is generally called interpreted (though you could argue that it is neither fish nor fowl).

However I have accepted that BASIC apparently doesn't work this way, I never knew.

Regards,

Tom
Edited 2020-06-24 10:10 by thwill

Game*Mite, CMM2 Welcome Tape, Creaky old text adventures

Turbo46

Guru

Joined: 24/12/2017
Location: Australia
Posts: 1593

Posted: 12:45am 24 Jun 2020

Copy link to clipboard

Print this post

Quote Files in not part of "core" MMBasic - it doesn't even exist on a MM2m, I'm talking about the main language elements. In general commands that can only be run at the command line are not "core"

It IS part of MMBasic for the original Mono Maximite, the Colour Maximite, the Micromite plus, DOS(Windows) MMBasic and probably other ported versions. It's part of Microsoft GWBASIC and QuickBASIC as well.

Sorry but I just think it's an unnecessary deviation from the well trodden path.

Bill

Keep safe. Live long and prosper.

Page 1 of 4

Print this page