[Gambas-user] Possible suggestion/question

[Emil Lenngren]

You can always hack the compiler/bytecode to be able to have gotos wherever
you want, but I think Gotos is not allowed to jump into control structures
because it is classed as "bad code structure". For ... Next also use hidden
local variables to store things.

But I think gosubs can be implemented in the compiler and only having to do
small changes to the interpreter, since using jumps/gotos to "return" into
control structures is not a problem because the hidden local control
variables can be guaranteed to be unchanged:
The compiler must analyze all the calls to a gosub and see how many control
structure variables that are occupied before entering it (the depth), so
the Gosub doesn't overwrite them.
One can use a control variable to store the return address of the Gosub, if
they are only allowed as leaf functions.

/Emil

2012/3/2 Jussi Lahtinen <jussi.lahtinen at ...626...>

> I made quick effort to try to simulate gosub with goto,
> and noticed that jumping back inside loops is not possible (as the
> documentation says...).
>
> I understand that when it's For ... Next, but shouldn't that be possible
> with Do ... Loop Until x?
>
> Anyway, I'm not sure my simulation does accurately compare speed of Sub to
> Gosub... if it does,
> then gosub would be 2 to 3 times faster.
>
> Jussi
>
> PS. Try to change last line "End" to "EndEnd" and then try to compile!
>
>
>
>
> 2012/3/2 Benoît Minisini <gambas at ...1...>
>
> > Le 25/02/2012 19:18, nando a écrit :
> > > Benoit,
> > >
> > > I came across the need to call many different common small pieces of
> > code multiple times.
> > > Naturally, we tend to make a SUB just for all of them.
> > >
> > > I wondered if you'd consider resurrecting GOSUB {label}
> > > which is in the same sub.
> > >
> > > It would be in the same scope, no parameters allowed and
> > > acts just like a leaf subroutine with only return on the stack
> > > and no formal stack frame.  It's also part compromise to embedded SUBs
> > >
> > > Would overal execution run faster ?
> > >
> > > ??
> > > -Fernando
> > >
> >
> > It may be a little bit faster. You can check that by simulate the GOSUB
> > with some GOTOs and compare the speed with a similar function call.
> >
> > Anyway, there is a syntax problem. In Basic, GOSUB is ended by a RETURN
> > instruction, but RETURN is already used to exit the all function. Or
> > maybe RETURN will return from the function only if there is no pending
> > GOSUB.
> >
> > Can you test if there is any speed gain I as said above? I don't have
> > the time to do that now. If the speed gain is worth it, then I may
> > implement GOSUB!
> >
> > Regards,
> >
> > --
> > Benoît Minisini
> >
> >
> >
> ------------------------------------------------------------------------------
> > Virtualization & Cloud Management Using Capacity Planning
> > Cloud computing makes use of virtualization - but cloud computing
> > also focuses on allowing computing to be delivered as a service.
> > http://www.accelacomm.com/jaw/sfnl/114/51521223/
> > _______________________________________________
> > Gambas-user mailing list
> > Gambas-user at lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/gambas-user
> >
>
>
> ------------------------------------------------------------------------------
> Virtualization & Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
> _______________________________________________
> Gambas-user mailing list
> Gambas-user at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/gambas-user
>
>