1. define_c_func/proc question
If define_c_func() is called several times for the same {dll,name} or
{{},address} pair, will the same routine_id be returned?
RC: could you include this precision in the docs? It's useful to know
when you want a custom message to call an user-supplied callback to pass the
results back.
Also, could you consider extending allocate() and free() to sequences? The
extended routines would return or free a sequence of addresses.
CChris
2. Re: define_c_func/proc question
- Posted by Robert Craig <rds at RapidEuphoria.com>
Dec 26, 2005
-
Last edited Dec 27, 2005
CChris wrote:
> If define_c_func() is called several times for the same {dll,name} or
> {{},address} pair, will the same routine_id be returned?
No, you'll get a different routine id each time.
In contrast, routine_id() will give you the same id
each time for the same Euphoria routine. In the case of
Euphoria routines, it was easy to check if a routine already
had an id assigned to it.
For most programs, it doesn't matter, because the routine id
is only derived once.
> RC: could you include this precision in the docs? It's useful to know
> when you want a custom message to call an user-supplied callback to pass the
>
> results back.
OK.
I'm always a bit leary of documenting minor details like this,
since it more or less commits me to never changing how it
works in the future, but I guess I can commit to this.
> Also, could you consider extending allocate() and free() to sequences? The
> extended routines would return or free a sequence of addresses.
I really don't see much demand for this.
It would be easy for you to make your own routines for this,
or just write a simple for-loop.
Regards,
Rob Craig
Rapid Deployment Software
http://www.RapidEuphoria.com
3. Re: define_c_func/proc question
- Posted by CChris <christian.cuvier at agriculture.gouv.fr>
Dec 29, 2005
-
Last edited Dec 30, 2005
Robert Craig wrote:
>
> CChris wrote:
> > If define_c_func() is called several times for the same {dll,name} or
> > {{},address} pair, will the same routine_id be returned?
>
> No, you'll get a different routine id each time.
>
> In contrast, routine_id() will give you the same id
> each time for the same Euphoria routine. In the case of
> Euphoria routines, it was easy to check if a routine already
> had an id assigned to it.
>
> For most programs, it doesn't matter, because the routine id
> is only derived once.
In the case at hand, the issue is with defining multiple ids for an user
defined routine whose address is passed to an EUphoria routine through a
custom Windows message. The Eu routine needs to go through this in order
to call the user back and send it the required data which may be a sequence
more complex than a plain string.
Checking the passed pair of arguments against an ever growing list of already
processsed calls doesn't appear to be very difficult or performance
penalising.
Or I missed something...
>
> > RC: could you include this precision in the docs? It's useful to know
> > when you want a custom message to call an user-supplied callback to pass the
> >
> > results back.
>
> OK.
> I'm always a bit leary of documenting minor details like this,
> since it more or less commits me to never changing how it
> works in the future, but I guess I can commit to this.
>
Just say that define_c_func() may return different ids. And you'll change
that the day where you'll be sure to return an unique id, which may never
come. Thys the coder will know s/he may have to handle different returned
ids, and hence will.
> > Also, could you consider extending allocate() and free() to sequences? The
> > extended routines would return or free a sequence of addresses.
>
> I really don't see much demand for this.
> It would be easy for you to make your own routines for this,
> or just write a simple for-loop.
>
Of course. Just a convenience, and more in line with most Eu routines
processing sequences transparently.
CChris
> Regards,
> Rob Craig
> Rapid Deployment Software
> <a href="http://www.RapidEuphoria.com">http://www.RapidEuphoria.com</a>
