From: "Samuel T. Harris" <sam_harris@hso.link.com>
Subject: Re: Can't export object of private type
Date: 1999/02/26
Date: 1999-02-26T00:00:00+00:00 [thread overview]
Message-ID: <36D6CC3E.7FFAFC10@hso.link.com> (raw)
In-Reply-To: 7b6f9k$at2$1@nnrp1.dejanews.com
bourguet@my-dejanews.com wrote:
>
> In article <7b5u87$s88$1@nnrp1.dejanews.com>,
> robert_dewar@my-dejanews.com wrote:
> ...
> > The advantage of private parts is of course the nice
> > compartmentalization. Indeed we have on the drawing boards
> > for GNAT a feature to have the private parts in separate
> > files which will be very handy for target specialization
> > and indeed for general abstraction.
> ...
>
> Well, this is a feature I'd like, especially if the compilation of
> a unit "withing" the package can be compiled (and code generated)
> without the private part. More or less like inter package inline
> is currently handled by gnat: you have a performance gain if you accept
> the dependancy but you may prefer the flexibility of the independancy...
> well I'd even accept that all the program have to be compiled with the
> same mode.
>
> -- Jean-Marc
I can say from experience with the old Rational R1000 that
such a facility is very powerful in dealing with variant code,
especially code which varies because of target architectures
to be supported.
While the R1000 did not allow putting the private part in
a separate file as, I believe, Robert is proposing, its
delination of spec views and code views kept the private
part in the code view (or at least that was the perceived
affect). Thus changes to the private part of a package
never caused a compilation hit on dependent units.
Using private or limited types is clearly advantageous
from a design level because dependent units have no
visibility to the private type declaration. However,
classical Ada compilers still have a compilation
dependency. One can work around this by defining the
private type as an access type to an incomplete type
definition in the private part. In this case, one is
allowed to complete the accessed type in the package
body. While this eliminates the compilation dependency,
it introduces a (sometimes unwanted) pointer.
This construction allows one to hide all the variant
details in the package body thus providing a single
specification for all variations. This is a very
important goal.
Allowing the private part in a separate also file gives
me the best of both. I avoid the recompilation hit and
I don't have to use access types if I don't want to.
I expect gnatchop will have an option to take existing
source and split out the private part as appropriate.
The reverse opertion will also be needed.
Will the default file extension be xxx.adp?
--
Samuel T. Harris, Principal Engineer
Raytheon, Scientific and Technical Systems
"If you can make it, We can fake it!"
next prev parent reply other threads:[~1999-02-26 0:00 UTC|newest]
Thread overview: 128+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <F7JoCB.JxB@syd.csa.com.au>
1999-02-24 0:00 ` Can't export object of private type Don Harrison
1999-02-24 0:00 ` Tom Moran
1999-02-24 0:00 ` Samuel Mize
1999-02-24 0:00 ` Tucker Taft
1999-02-25 0:00 ` Don Harrison
1999-02-25 0:00 ` robert_dewar
1999-02-26 0:00 ` Don Harrison
1999-02-26 0:00 ` robert_dewar
1999-02-26 0:00 ` dennison
1999-02-26 0:00 ` bourguet
1999-02-26 0:00 ` Samuel T. Harris [this message]
1999-02-27 0:00 ` Simon Wright
1999-02-27 0:00 ` Jean-Pierre Rosen
1999-02-28 0:00 ` dewar
1999-03-01 0:00 ` bourguet
1999-03-01 0:00 ` Don Harrison
1999-03-01 0:00 ` robert_dewar
1999-03-03 0:00 ` Don Harrison
1999-03-03 0:00 ` robert_dewar
1999-03-01 0:00 ` Stephen Leake
1999-02-27 0:00 ` Brian Rogoff
1999-03-01 0:00 ` robert_dewar
1999-02-25 0:00 ` Samuel Mize
1999-02-26 0:00 ` Don Harrison
1999-02-27 0:00 ` Nick Roberts
1999-03-01 0:00 ` Don Harrison
1999-03-02 0:00 ` Matthew Heaney
1999-03-01 0:00 ` Don Harrison
1999-03-01 0:00 ` Matthew Heaney
1999-03-01 0:00 ` Nick Roberts
1999-03-01 0:00 ` Nick Roberts
1999-03-01 0:00 ` Don Harrison
1999-03-02 0:00 ` Matthew Heaney
1999-03-03 0:00 ` Don Harrison
1999-03-03 0:00 ` Samuel Mize
1999-03-04 0:00 ` Don Harrison
1999-03-07 0:00 ` Ehud Lamm
1999-03-03 0:00 ` Robert A Duff
1999-03-04 0:00 ` Don Harrison
1999-03-04 0:00 ` Robert A Duff
1999-02-28 0:00 ` Matthew Heaney
1999-03-01 0:00 ` Nick Roberts
1999-03-01 0:00 ` Matthew Heaney
1999-03-01 0:00 ` Matthew Heaney
1999-03-02 0:00 ` Nick Roberts
1999-03-01 0:00 ` Samuel Mize
1999-02-25 0:00 ` fraser
1999-02-26 0:00 ` Don Harrison
[not found] ` <7b6nqe$75m$1@remarq.com>
1999-02-26 0:00 ` fraser
1999-02-27 0:00 ` Nick Roberts
1999-02-26 0:00 ` fraser
1999-03-01 0:00 ` Don Harrison
1999-03-01 0:00 ` Matthew Heaney
1999-02-26 0:00 ` Samuel Mize
1999-03-01 0:00 ` Don Harrison
1999-03-01 0:00 ` Matthew Heaney
1999-03-02 0:00 ` fraser
1999-03-03 0:00 ` Don Harrison
1999-02-28 0:00 ` Matthew Heaney
1999-02-28 0:00 ` Matthew Heaney
1999-02-28 0:00 ` Matthew Heaney
1999-03-01 0:00 ` Tom Moran
1999-03-02 0:00 ` Matthew Heaney
1999-03-02 0:00 ` Tom Moran
1999-03-02 0:00 ` Matthew Heaney
1999-03-02 0:00 ` Tom Moran
1999-03-02 0:00 ` Matthew Heaney
1999-03-02 0:00 ` nabbasi
1999-03-02 0:00 ` Matthew Heaney
1999-03-03 0:00 ` Don Harrison
1999-03-03 0:00 ` Single Extension; Polymorphic Arrays Nick Roberts
1999-03-03 0:00 ` Nick Roberts
1999-03-08 0:00 ` Matthew Heaney
1999-03-08 0:00 ` Nick Roberts
1999-03-08 0:00 ` Tucker Taft
[not found] ` <m3ogm40wav.fsf@mheaney.ni.net>
1999-03-08 0:00 ` Nick Roberts
1999-03-08 0:00 ` Tucker Taft
1999-03-08 0:00 ` dennison
1999-03-09 0:00 ` Nick Roberts
1999-03-03 0:00 ` Can't export object of private type Don Harrison
1999-03-03 0:00 ` Don Harrison
1999-03-03 0:00 ` Nick Roberts
1999-03-04 0:00 ` Don Harrison
1999-03-04 0:00 ` Nick Roberts
1999-03-04 0:00 ` robert_dewar
1999-03-05 0:00 ` Nick Roberts
1999-03-05 0:00 ` Robert A Duff
1999-03-05 0:00 ` Abstract Subprograms of Untagged Types Nick Roberts
1999-03-05 0:00 ` robert_dewar
1999-03-05 0:00 ` Tucker Taft
1999-03-05 0:00 ` Nick Roberts
1999-03-06 0:00 ` robert_dewar
1999-03-04 0:00 ` Can't export object of private type Nick Roberts
1999-03-08 0:00 ` Matthew Heaney
1999-03-09 0:00 ` Don Harrison
1999-03-09 0:00 ` Matthew Heaney
1999-03-09 0:00 ` Nick Roberts
1999-03-10 0:00 ` Don Harrison
1999-03-10 0:00 ` Matthew Heaney
1999-03-04 0:00 ` fraser
1999-03-09 0:00 ` Don Harrison
1999-03-08 0:00 ` Matthew Heaney
1999-03-08 0:00 ` Nick Roberts
1999-03-08 0:00 ` Matthew Heaney
1999-03-10 0:00 ` Don Harrison
1999-03-10 0:00 ` Matthew Heaney
1999-03-10 0:00 ` dennison
1999-03-10 0:00 ` robert_dewar
1999-03-10 0:00 ` dennison
1999-03-10 0:00 ` robert_dewar
1999-03-10 0:00 ` dennison
1999-03-11 0:00 ` robert_dewar
1999-03-11 0:00 ` Don Harrison
1999-03-12 0:00 ` robert_dewar
1999-03-11 0:00 ` bill
1999-03-11 0:00 ` dennison
1999-03-11 0:00 ` Scott Ingram
1999-03-11 0:00 ` Larry Kilgallen
1999-03-12 0:00 ` dewar
1999-03-11 0:00 ` dennison
1999-03-11 0:00 ` Don Harrison
1999-03-10 0:00 ` Robert A Duff
1999-03-10 0:00 ` dennison
1999-03-11 0:00 ` dennison
1999-03-10 0:00 ` robert_dewar
1999-03-03 0:00 ` Don Harrison
1999-02-28 0:00 ` Matthew Heaney
1999-02-28 0:00 ` Matthew Heaney
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox