OpenEuphoria

2. Re: load user defined module

• Posted by Sabal.Mike at notations.com Nov 07, 2002
• 412 views

Correct, this won't work because include must be a top level routine
with a static filename.  One oft-proposed solution is to open an include
file, puts(fn,"include "&x&"\n"), close fn, then include the include
file.  It's not a very elegant solution, but it should do what you need
it do.

I've proposed before, and in light of current discussions will do so
again now, that some sort of overlay mechanism is needed.  I should be
able to launch a variable Euphoria program from within a Euphoria
program without executing a new shell or console.  This, IMHO, would be
a much more valuable addition than certain other features that I never
use.

Mike Sabal

>>> salix at freemail.hu 11/07/02 10:13AM >>>

include get.e
object z
z=prompt_string("What to load? ")
procedure inc(sequence x)
	include x
end procedure
inc(z)

-- This doesn't work
-- But how to get around it?
-- Cheers, Salix 

4. Re: load user defined module

• Posted by Derek Parnell <ddparnell at bigpond.com> Nov 07, 2002
• 394 views

To: Robert Craig

Rob, how would you recommend that coders deal with this issue of
platform-specific code, using the
current incarnation of Euphoria?



8/11/2002 11:46:30 AM, stabmaster_ at hotmail.com wrote:

>
>Personally, I'd like to see something like this:
>
>ifplatform DOS32
>-- DOS-specific code and includes goes here
>endif
>
>ifplatform LINUX
>-- and so on...
>
>
>ifplatform should be checked by Ex/Exw/Exu as a first pass, only reading 
>code for the given platform (since the current platform is known by the 
>interpreter at all times).
>
>
>
>
---------
Cheers,
Derek Parnell

6. Re: load user defined module

• Posted by stabmaster_ at hotmail.com Nov 08, 2002
• 407 views

>Euphoria does not have conditional compilation built into it,
>like C does. I'd like to keep it that way. Why add
>a new concept to the language, for such a small advantage?

I see it as a bit more than a "small" advantage. Since Euphoria is a 
multi-platform language, and hopefully will be ported to even more platforms 
in the future, conditional interpretation would imo be the most 
straightforward way to write programs that will run on all supported 
platforms. Sure, writing include-statements to a bogus file and then 
including it works. But it's not really intuitive, it kind of a trick that 
just so happens to work.

8. Re: load user defined module

• Posted by Derek Parnell <ddparnell at bigpond.com> Nov 08, 2002
• 385 views

Ooops, I clicked send before I could add ...

Now consider the possibility of a Palm OS version of Euphoria. Using the
current platform() method, we now have a lot of work to do to port it.
(Assuming that a palm/ui.e was already defined). Whereas the idea of an
include being used inside an IF would be a trival exercise.

----------------
cheers,
Derek Parnell

10. Re: load user defined module

• Posted by petelomax at blueyonder.co.uk Nov 08, 2002
• 391 views

{{{ On Sat, 9 Nov 2002 00:26:38 +1100, Derek Parnell <ddparnell at bigpond.com> wrote:

app.ex: include dos/ui.e include main.e

app.exw: include win32/ui.e include main.e

app.exu: include linux/ui.e include main.e

Providing the three variants of ui.e have a consistent api which you are suggesting then main.e need only contain platform independent code & no calls to platform() either.

I think it would be a mistake to actually include say win32lib.ew when running on say linux where it is never going to get called.

Pete

12. Re: load user defined module

• Posted by Robert Craig <rds at RapidEuphoria.com> Nov 08, 2002
• 389 views

Derek Parnell writes:
> Robert,
> is this the sort of thing you are suggesting for us ? ...
> ...
> ...
> As you could imagine, it would not take much for this to
> become very hard to maintain. Can you suggest a better approach?

I don't know. There are various approaches that you could take.
Igor's dynamic include approach is one, although it won't
work with the binder or translator.
A lot depends on the specific situation - how much code is in
common across platforms, can the differences be expressed as
different data rather than different code, etc. You might even
use 3 different sequences of routine id's for 3 platforms.
I'm sure you can come up with as good a solution as I might.
The conditional compilation approach can get ugly too.
If you look at the C source for Euphoria, you'll see sections of
very ugly code because of all the conditional compilation
directives for different platforms. I do it because people
demand maximum speed and minimum interpreter size.
I'd need to see some real cross-platform examples 
to get interested in adding conditional compilation.
I just haven't seen many.

Regards,
   Rob Craig
   Rapid Deployment Software
   http://www.RapidEuphoria.com

14. Re: load user defined module

• Posted by Juergen Luethje <jluethje at gmx.de> Nov 09, 2002
• 419 views

Rob wrote:

<snip>
> I'd need to see some real cross-platform examples
> to get interested in adding conditional compilation.
> I just haven't seen many.

Sorry, I don't understand. What do you mean with
"real cross-platform examples"?

TIA and regards,
   Juergen

-- 
while not asleep do
   sheep += 1
end while

16. Re: load user defined module

• Posted by Robert Craig <rds at RapidEuphoria.com> Nov 09, 2002
• 401 views

Juergen writes:
> Sorry, I don't understand. What do you mean with
> "real cross-platform examples"?

There are lots of Euphoria programs and libraries that run
on multiple platforms. e.g. ed.ex, sort() etc.
Sometimes there's a need to use platform() to
select different statements for different platforms, e.g
    if platform() = LINUX then
         system("rm temp", 2)
    else
         system("del temp", 2)
    end if

In the examples that I've seen so far, the platform-dependent
sections are usually fairly small, and only form a small 
percentage of the code. I suppose what I meant by
the fuzzy term "real cross-platform example", was 
a large program or library with many big chunks of code
that are different on the different platforms. You might imagine
something like a combined Windows/Linux version of Win32Lib, 
except that the differences in that case would be so great 
that you'd be better off maintaining two completely separate 
win32lib.ew files, rather than using platform() all over the place.
There are probably some better examples, where it would make
sense to maintain one source with heavy use of platform(),
but I was just saying that I haven't seen those examples yet,
so I don't see an urgent need for better support in this
area, such as C-style conditional compilation.

Regards,
   Rob Craig
   Rapid Deployment Software
   http://www.RapidEuphoria.com