OpenEuphoria

Re: The fate of Euphoria=20

• Posted by Tone Škoda <tskoda at email.si> Nov 11, 2004
• 412 views

Patrick Barnes wrote:
> 
> On Wed, 10 Nov 2004 13:02:50 -0800, Tone =C5=A0koda <guest at
> rapideuphoria.co=
> m> wrote:
> > posted by: Tone =C5=A0koda <tskoda at email.si>
> >
> > Patrick Barnes wrote:
> >
> > > The features that I feel Euphoria is 'missing' are
> > > better flow control - continue, continue and exit with specified level.
> >
> > I don't find these so neccessary.
> 
> That's alright, perhaps there is something you feel would be nice to
> have in Euphoria, but I am rather ambivalent about.
> 
> > > the ability to give sequences a rigid structure when it's needed.
> >
> > There should be something done about this, but I don't have a good sugges=
> tion what.=20
> > Maybe it's best if it stays as it is, using types.
> 
> Well, that's what the 'of' specification is. An enhancement to the
> types system that makes them much faster, and more capable.

with type_check is sometimes really very slow. I don't know why. I have
EXTRA_DEBUG variable, and then i do this:

if EXTRA_DEBUG then


> 
> > I use "structures" a lot, in fact all data is saved in structured sequenc=
> es.
> 
> Me too. In fact, no code I have ever seen has a large sequence in the
> program that is *NOT* structured.
> 
> > > Did you see the document I sent to the list back in June about the
> > > 'of' statement? It would make structures possible and type checking
> > > much, much faster (As tests have proved)
> > > <a href="http://users.secsme.org.au/~prbarnes/misc/oth/type_of.txt">h=
> ttp://users.secsme.org.au/~prbarnes/misc/oth/type_of.txt</a>
> >
> > At first look it looks a bit complicated.
> 
> Well, I'm not the *best* technical writer. However, it's rather
> in-depth because it contains all data necessary to implement the idea,
> look here: <a
> href="http://users.secsme.org.au/~prbarnes/misc/oth/of.zip">http://users.secsme.org.au/~prbarnes/misc/oth/of.zip</a>
> 
> For the programmer using it, it's very simple... Contrived example:
> 
> Old type:
> type int_array( object x )
>      if sequence(x) then
>            for a = 1 to length(x) do
>                  if not integer(x[a]) then
>                        return 0
>                  end if
>            end for
>            return 1
>      end if
>       return 0
> end type
> 
> New type:
> type int_array( sequence of integer x)
>      return 1
> end type
> 
> }}}
<eucode>
> int_array test_array
> test_array = {}
> for a = 1 to 100 do
>     test_array &= rand(100)    --Add an integer to the end of test_array
> end for
> test_array &= 1.05   --The program will crash here, due to a type_check f=
> ailure.
> </eucode>
{{{

> 
> The advantage over implementing a traditional 'type' that does the
> same thing, is that to get through the for-loop above, the traditional
> type has to check 5500 (1+2+3+4...+10) elements, whereas the new 'of'
> type only has to check 100.
> 
> If speed was sorely needed for a tight loop, then perhaps a
> type_check([0|1]) function could be added, to turn type checking on
> and off dynamically.
> 
> --
> MrTrick
> ----------
> 
>