Re: Modes (was unbounded_string)

["Jeff C," <nolongersafeto@userealemailsniff.com>]

"Warren W. Gay VE3WWG" <ve3wwg_nospam@cogeco.ca> wrote in message
news:Mb2jb.8531> Well, C returns a pointer, if you use getenv(2). However,
there is no reason
> that ACT could not have converted this into a normal Ada string, which is
> what I would have expected. Even in C, you are really expected to use
> putenv or setenv to make a change. However, programmers have been
> known to "cheat" and modify the string that the getenv pointer points to,
> which works provided that the string does not require more length. But
> even in C terms, IMHO, this is a bad practice.
>


Never assume that there is "no reason" for ACT or anyone else for that
matter doing something without
at least quickly looking at it. The start of the package says

-- This package tends to use fairly low-level Ada in order to not bring
-- in large portions of the RTL. For example, functions return access
-- to string as part of avoiding functions returning unconstrained types.


gnat.os_lib is useful but it probably should not be generally used within
you code. It certainly would
be nice if ACT provided a "thicker" version of the package (and perhaps if a
few customers ask
for it they will) but it appears that they had their reasons whey they
originally set this up.