Re: How to get Ada to ?cross the chasm??

[Marius Amado-Alves <amado.alves@gmail.com>]

On Friday, 11 May 2018 13:46:34 UTC+1, Lucretia  wrote:
> On Thursday, 10 May 2018 23:52:18 UTC+1, Paul Rubin  wrote:
> > "Randy Brukardt" <randy@ com> writes:
> > > No wonder you don't understand: Ada 95 had none of the container stuff or 
> > > other features that reduces the memory management burden without using GC. 
> > > At a minimum, you need to consider Ada 2005, but the most recent Ada is 
> > > best.
> > 
> > Ah, ok, I was also going by
> > 
> >   http://cowlark.com/2014-04-27-ada/index.html
> 
> I wouldn't, that uses some weird syntax, it's like he's a C programmer doing Ada.
> 
> > which claims Ada has no standardized way to release allocated memory (so
> > each implementation had its own way).  If you can suggest a good place
> > to read about the current standard method I'd appreciate it.
> 
> Bullshit.
> 
> http://www.ada-auth.org/standards/rm12_w_tc1/html/RM-13-11-2.html
>  
> > To be fair, the Pascal-like subset I looked into first was SPARK (I read
> > John Barnes' book about it) since SPARK was the aspect of Ada that at
> > first interested me the most.  I'd be interested to know how the Ada
> > 2012 containers differ from C++ STL containers.  As mentioned elsewhere
> > I believe that the same guy (Alex Stepanov) had a hand in the design of
> > both.
> 
> Your previous post about him says he worked on *a* container library for Ada83 (it was in 1989) and he even says he never touched Ada much since then. So, no, he didn't have a hand in the current implementation.
> 
> I think it was Matthew Heaney who provided them, https://groups.google.com/forum/#!topic/comp.lang.ada/luv71Ex0CDc to the ARG for standardising, IIRC.
> 
> Luke.

Unfortunately, Ada.Containers is greatly modelled on the STL, yes. But it's nevertheless of excellent quality. Matthew Heaney provided the bulk of a prototype implementation that most probably has been integrated into the current GNATs.

My understanding of what happen in the 1980's: Alex Stepanov was looking for, and into, generic programming, and the only language providing that was Ada 83, and he looked briefly into her, but then for some reason he teamed up with Stoutstroup for generics in C++, which they did the C way, with preprocessing, hence "template". So the strange history of Ada.Containers is Ada 83 -> STL -> Ada 2005.

The goal in the 1980's was to make the data structure algorithms (linked lists, etc.) generic on the element type. Ada 83 was excellent for that, but the opportunity was waisted for some reason. Also, they let the algorithmic aspect show up more than it should, in the STL and then in Ada.Containers. Instead of the algebraic structure. All that fussing with null values, etc. This was the unfortunate part of Ada.Containers being modelled on the STL.