RTE under basic program ..pascal ,gfa ,omikron,basic

[christos]

Sorry i could not help you more.

Don't be so sad. It's not your fault you couldn't help

[charles]

of course he couldn't help ,
have you seen that bludgering he did

neat but destructive

unstable
and does not run under the editor
can not be reconfigured to work with other timers

charles

[spiny]

of course he couldn't help ,
have you seen that bludgering he did

neat but destructive

unstable
and does not run under the editor
can not be reconfigured to work with other timers

charles

irony meter at critical !

[Tomchi]

well as soon as i posted the code, it was meant to be used, modifyied by anyone, so do what you want with it, whatever moderized may mean even if it's to say my code doesn't work or yours is better.

Now don't say my code is unstable, coz the timer b example is just correct and very stable, now what bludgering mean i don't know, does not run in editor is just common thing as far as you code decent speed things with GfA overmore when GfA wasn't designed for the thing, cannot be configured to work with other timers just involves yourself .

the post asked for a way to make an RTE in a basic language, i showed you what i do ... Never spoke about being XX years experimenting things, now it doesn't work at first try for your purpose but still it writes an RTE , so please charles, don't pee on my code coz others are kidding you and you feel being better than me, coz you're far from that, thx.

[charles]

don't get me wrong
iam not peeing on your code
i am just making a point that a code which won't run in the editor is close to poo poo for my uses and trivial practise, how is this practical to design a program if users have to stop the editor , load the compilliier and compille
every time they want to see if the code thay are working on is "proofed right" ?

any how it doesn't work correctily and is a unstable method as i illustrated in my example , it only works for the installed timerb vector and is non transferable to the timer a vector
this means something is wrong if the same technque can not be applyed twice

i'll check again . but i'm afraid i do not like or really approve of hacks like this (frustrating to not see at edit time whats going on)

and yes like you said this is just one of many ways to do this
i'll find more!!!!!!!!!!!!!!!!!

thanks for work tomchi
charles

[christos]

don't get me wrong
iam not peeing on your code
i am just making a point that a code which won't run in the editor is close to poo poo for my uses and trivial practise, how is this practical to design a program if users have to stop the editor , load the compilliier and compille

charles

You realise that you don't really have to quit the editor with gfa 3.6TT. There's the menu.prg thingy..

[lp]

don't get me wrong
iam not peeing on your code
i am just making a point that a code which won't run in the editor is close to poo poo for my uses and trivial practise, how is this practical to design a program if users have to stop the editor , load the compilliier and compille
every time they want to see if the code thay are working on is "proofed right" ?

Calling it poo poo is like saying its sh*t. You probably burned another bridge.

You didn't ask for a solution that worked in the editor, you ask for a solution which left the door open for any solution. Then you get one, can't make it work, now its Tomchi's fault. Its actually your fault.

If you are having to quit the editor, reload the source by hand every time you need to edit it, then obviously your development environment sucks. There are shells and multi-tasking operating systems that ease this process. Again this is your problem.

i'll check again . but i'm afraid i do not like or really approve of hacks like this (frustrating to not see at edit time whats going on)

Then you should of stated you wanted a solution that works in the editor from the get go. As if its any worse than the peeking and poking you do all over your code that we have seen in the past.

and yes like you said this is just one of many ways to do this
i'll find more!!!!!!!!!!!!!!!!!

There's really only 2 ways, the method Tomchi uses or write the interrupt in asm and install it. Apparently you are trying to avoid the assembler. You were also told BASIC was not designed for this sort of thing.

[wongck]

now what bludgering mean i don't know.

Right... Google define seems to also run out of ideas of what it means...

Watching at the side lines is such fun with Charles in the match.

[charles]

thanks all clear in a cloudy sort of way

charles

[wongck]

thanks all clear in a cloudy sort of way
charles

Guess the reason is that it's snowy in Canada.

[charles]

Tomchi wrote:
now what bludgering mean i don't know

that implies decompostion

like saying " whats 1 and 1 "

and you say its "5 - 3"

we are both correct but asm or pure asm or even machine is more correct to a basic


so no object can envoke RTE?

[lp]

I think he ate the yellow snow.

[charles]

don't set me into passive mode
nothing will get written

charles

[Nyh]

Code: Select all

    PROCEDURE timer_b
      ~1                                          ! to locate start of procedure
      ~1                                          ! and to be replaced by the save registers machine code
      ' interrupt code
      CARD{&HFFFF8240}=CARD{rst%}                 ! rasterbar
      CARD{&HFFFF825E}=CARD{rst15%}               ! colour 15 rasters
      ADD rst%,2                                  ! points to next
      ADD rst15%,2                                ! points to next
      '
      BYTE{&HFFFFFA0F}=BCLR(BYTE{&HFFFFFA0F},0)   ! interupt is done
      '
      ~2                                          ! to locate end of procedure
      ~2                                          ! and to be replaced by the restore registers machine code
    RETURN


Now this is some naughty code. This will indeed work for rasters. It will however not work for timer interrupts because they might happen during a VBL interrupt (and a raster interrupt won't).

Simple advice to Charles: write your interrupt routines in assembly. There is no simple solution possible in Basic or Pascal. Having the RTE command is only a little part of the problem of writing an interrupt routine in Basic.

Hans Wessels

[charles]

nyh just optimized my day!! thanks
this explaination is what we were looking for

1+1 =5-3

i told tomchi my code wasn't suitable for my application

and other frowned upon this
especially the moderator
who got easilly bent out of shape

wheres mine ?

whats up?

charles

[charles]

so basically i have to add a little more code to my dervirved tomchis ,
like if vbl interupt happening

do not read my variable?



is this suggestable?

charles

[Tomchi]

Now this is some naughty code. This will indeed work for rasters. It will however not work for timer interrupts because they might happen during a VBL interrupt (and a raster interrupt won't).


Hans Wessels

I don't understand why it doesn't work. I have something working with timer A, so i thought VBL wasn't a problem, but then moving mouse or even doing a PRINT would crash the proggy.

Could you enlighhten me Nyh plz, about what happens if the interrupt occurs during a VBL and why asm would work and converted gfa won't ?

[charles]

here is my 2 cents

its interupt driven
the interupt is set off pulse , pulse is from a timer
if something else is refreshing under the interupt ,(at the same time) the forced action interferrs with installed action
and clogs the system, delay cpu execution,and
causing misfired exceptions
result "bomby" the sequel no not from disney!!!!!

but to write all this out in a more technical code explanation
i would be here long than some people like

charles

[Nyh]

so basically i have to add a little more code to my dervirved tomchis ,
like if vbl interupt happening

do not read my variable?

Yes indeed!

It is that simple!

The only thing you have to do is snatch the SR from the stack, read the interrupt mask bits and proceed if there is no VBL happening which only can be done in assembly.

We will completely ignore the fact that writing the interrupt routine is assembly is far easier and that you are completely incapable of writing such a framework in Gfa Basic. Because that doesn't matter.

Hans Wessels

[Nyh]

Now this is some naughty code. This will indeed work for rasters. It will however not work for timer interrupts because they might happen during a VBL interrupt (and a raster interrupt won't).

Hans Wessels

I don't understand why it doesn't work. I have something working with timer A, so i thought VBL wasn't a problem, but then moving mouse or even doing a PRINT would crash the proggy.

Could you enlighhten me Nyh plz, about what happens if the interrupt occurs during a VBL and why asm would work and converted gfa won't ?

All variables in Gfa are referenced A5 relative (if I am not mistaken). When a VBL or mouse interrupt in interrupted by an other interrupt like your timer interrupt it just might be A5 is pointing to something completely else and reading or writing to A5 will have interesting effects which will result in bombs. Even during a print statement A5 isn't pointing where it is supposed to point during execution of compiled Gfa code.

There are 2 solutions to this problem:
1 don't touch any variables, use only pokes and peeks like
wpoke $ff8240,wpeek(lpeek($fc))
lpoke $fc,lpeek($fc)+2
$fc contains a pointer to your rasterbar table.
All unused exception vectors can be misused as variables for interrupt routines. Depending on the Gfa compiler there might be no need to save any register because the compiler will not use any registers. I am pretty sure the Omikron compiler will not use any registers for this code but I don't know the Gfa compiler that good.

2 set A5 to the correct value so you can touch variables. The right value for A5 has to be stored at an address that can be reached by the assembly at the start of the interrupt routine. This can not be done in Gfa basic.

Hans Wessels

[lp]

don't set me into passive mode
nothing will get written

You've been in passive mode for 9+ years already. I tried to tell you some posts back that BASIC is not suitable for this, but you don't listen. Typical.

Its more than a5, GFA maintains registers a3-a5 always. If any of these change, GFA fails. Its stated in the manual which charles probably never fully read. It applies to the editor and the compiler. They work nearly identical.

[Nyh]

Its more than a5, GFA maintains registers a3-a5 always. If any of these change, GFA fails. Its stated in the manual which he probably never fully read. It applies to the editor and the compiler. They work nearly identical.

I stand corrected. Thanks for the clarification. And yes, I didn't read the Gfa manual form front to back because, well, I never did do any programming in Gfa.

Hans Wessels

[Nyh]

Stupid me. I did a double post.

[lp]

Its more than a5, GFA maintains registers a3-a5 always. If any of these change, GFA fails. Its stated in the manual which he probably never fully read. It applies to the editor and the compiler. They work nearly identical.

I stand corrected. Thanks for the clarification. And yes, I didn't read the Gfa manual form front to back because, well, I never did do any programming in Gfa.

he=charles

I meant charles, not you. Charles is the one trying to make GFA do things it was never meant to do, therefore its his responsibility to read the manual. I have gone back and corrected the post.

[simonsunnyboy]

*Mod note* Locked as there is nothing more to say. This hacky technique works but its limitations have been made clear.

In short, BASIC on it's own was never designed nor meant to handle true hardware interrupts (be it timers or other hardware).

Generic solution: Ensure the problem to be solved really needs an interrupt routine and if it does, write this interrupt handler in assembly language or ask someone to do this for you.