From: Maciej Sobczak <see.my.homepage@gmail.com>
Subject: Re: Ravenscar and portability
Date: Fri, 31 Aug 2007 00:23:06 -0700
Date: 2007-08-31T00:23:06-07:00 [thread overview]
Message-ID: <1188544986.420853.189520@z24g2000prh.googlegroups.com> (raw)
In-Reply-To: <wcc1wdkipst.fsf@shell01.TheWorld.com>
On 31 Sie, 01:08, Robert A Duff <bobd...@shell01.TheWorld.com> wrote:
> > The Ravenscar profile contains the No_Implicit_Heap_Allocations
> > constraint.
>
> > The problem is that the operations that might require implicit heap
> > allocation are implementation-defined. This means that programs can be
> > declared as Ravenscar-compliant *only* in the context of some chosen
> > Ada implementation.
> I believe all of Ada's features, excluding some predefined library
> units, can be implemented without implicit heap allocation.
What about unconstrained types? Indeed, basic use cases can be
implemented on the stack, but I might be missing something less
obvious.
> But I'm not even sure how to precisely define "implicit heap
> allocation".
Good question, but I was not around when Ravenscar was defined...
> The Strings.Unbounded does heap allocation. Is it "implicit"?
I would say yes, because the usage of heap is not stated in the
standard. It might be a common implementation practice and there might
even be no other choice, but it is not stated.
--
Maciej Sobczak
http://www.msobczak.com/
prev parent reply other threads:[~2007-08-31 7:23 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-08-29 7:52 Ravenscar and portability Maciej Sobczak
2007-08-30 23:08 ` Robert A Duff
2007-08-31 7:23 ` Maciej Sobczak [this message]
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox