From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,404527a85668e8ac
X-Google-Attributes: gid103376,public
From: Al Johnston <sofeise@mindspring.com>
Subject: Re: Single statment STRING to chars_ptr... how to?
Date: 2000/03/15
Message-ID: <38CFD170.C8979F1D@mindspring.com>#1/1
X-Deja-AN: 597897314
Content-Transfer-Encoding: 7bit
References: <38CE7BB6.C045C710@mindspring.com> <8ao96a$nk0$1@nnrp1.deja.com>
To: dennison@telepath.com
X-Accept-Language: en
X-Server-Date: 15 Mar 2000 18:09:49 GMT
Content-Type: text/plain; charset=us-ascii
Organization: MindSpring Enterprises
Mime-Version: 1.0
Newsgroups: comp.lang.ada
Date: 2000-03-15T18:09:49+00:00
List-Id: <comp.lang.ada>

THANK YOU...  I have been pounding my head against this for a day
now...

> The problem is that you are trying to create a pointer to a stack-based
> copy of my_string (specificly, the copy passed by by interfaces.c.to_c).
> That copy in all likelyhood ceases to exist as a valid object after the

well, when you put it that way... :{

two quick points.  1) I am new to 95 and am simply trying to learn
what to do and what not to do.  2) Our real code (being ported
from 83 to 95 by me) uses system.address for all of the c interfaces.
Some of these probably will work under gnat, some of them will not
work (fat pointers), and some I just dont know about.

In the case you responded to I am trying to find a replacement for a
a single statement in our 83 code, preferably without having to write any
subroutines of my own.  Here is what the code looks like under 83:

    return mypkg.my_c_sub(mypkg.cstring_to_adastring(the_string));

where my_c_sub is written in c and imported, and the_string is
just a normal ada string.

a second case is

    mypkg.his_c_sub(mypkg.cstring_to_adastring(the_string));
For both of these cases I want the c routines (my_c_sub and
his_c_sub) to have a "in" parameter of type
interfaces.c.strings.chars_ptr;

I know I am trying to have a pointer to memory on the stack, but
I was expecting the compiler to be able to carry this construct
back up the function returns.  In the particular case of

    if.c.s.to_chars_ptr[if.c.to-c(S)'access]

my intent was to have if.c.to-c(S) to return its value by
VALUE, via the stack and have the outer function
if.c.s.to_chars_ptr operate on that value, making a new
thingie on the stack that it could then return by VALUE
to its caller, which would then be still "living" in the
scope of the caller. This type of thing is done all the time
with arrays and structures... so I thought it could be done
with some type of anonymous pointer type that would
be created/distroyed as needed to propagate the data
up the call list.

the think on the stack that is pointed to is only there
for the interface between two routines, so it lifetime
should be okay... I thought I had seen code that did
this type of thing.

Well, that's what I was trying to do, and why I thought it would
be okay... so, where am I going awry?

By the way, the reason I was trying to use chars_ptr was I was
familiar with it.  As best as I can tell, using if.c.char_array is
actually a better choice for the two "real code" cases I gave
above.  Still, I am try to learn the can's and cant's of chars_ptr,
so I am still very interested in this.


thanks for the help!

-al