Re: Anybody in US using ADA ? New language competition? (long)

["Warren W. Gay VE3WWG" <ve3wwg@cogeco.ca>]

Marin David Condic wrote:
> Obviously, you can't speak for all vendors, but you could speak for at least
> *one* vendor. :-) (Probably, you also talk to most of the other ones with
> your involvement in other areas of Ada.) Would there be any sympathy on the
> part of the vendors (or RR Software?) to get some sort of "Conventional Ada
> Library" going under the auspices of SIGAda? Would there be some willingness
> on the part of the vendors (or RR Software) to distribute such a library?
> 
> SIGAda probably would be a good place to get some kind of development
> package put together from things on the net. 

My only concern with SIGAda being the only one to distribute such a
beast is that their interest may be a bit different than many
Linux/*BSD or even Win32 enthusiasts have in mind. If they focus
too much on making it work for all of the commonly available Ada
compilers out there, I can see how this might actually hold back
such a project.

Whereas, if you focused on the hobby community (at least for a start),
where they are already doing major things like writing O/S's (Linux,
Hurd, etc.), X-Windows (updates), LessTif (MOTIF replacement) etc.,
then you might get more enthusiasm and overall support over time.
This lets you leverage the GNAT features, like some of the GNAT
supplied libraries (although, you could probably still distribute
those libraries for use with other compilers, if you are willing
to make the necessary changes).

The other important factor I see in a successful distribution is that
certain elements of it should be _optional_ to install.  If you
have biases against certain packages and want to insure that none
of yours will use them for example, the best way to prevent that
is to not waste space installing them. OTOH, this may make it
harder to install some other Ada software that needs it, but hey--
empower the end user anyway. ;-)  Disk space limitations may be
one of the most compelling reasons why not all packages will
be selected for install.

Doing this, also mandates some sort of a package dependency
check. All of this type of thing is being done already in Linux
distributions, and I would like to see a similar result.

(Blue Sky Warning -- Visors down)  It would be truly wonderful to
be able to do a "Ada Package Update" from the net, dare I say it,
in a manner similar to the "Windows Update" or "RedHat update".
This would be a service worth paying for.

> Selecting some things (like a
> database product?) would be difficult because you'd want it to work on
> several platforms and everyone is going to have a favorite, but it might be
> possible to do that relatively quickly.

It doesn't have to be difficult. Use different packages for different
client support. For example, you could have my APQ for PostgreSQL
package installed, whether you plan to use it or not (some other
Ada Open Sourced project might ;-)  You could have another package,
maybe named My_SQL that could exist along side. Again, if you
can deselect either at install/update time, you never need to be
forced to have both, but they can sit there too, if you like.

> (Version 1 doesn't have to have
> *everything* in it - just a good start.) 

Absolutely.

> You're right about it being work to
> make sure it compiles across most of the compilers out there, but
> realistically, its a job that could get done if there were enough volunteers
> willing to take a package and run it against what they've got. 

I think if a GNAT version were released, vendors would have their own
invested interest in getting the same distribution working for their
own products. I think this could kind of pressure would bring about
the desired side effect ;-)  Then you might see such a beast coming
with your favour compiler update, from the vendor.

> If a disk
> were produced, SIGAda might even be able to raise some funds by selling it
> for some small fee.

Because it adds so much value, this I believe would work.

> But again, I think it would have a better chance of success if the vendors
> were to get behind it and at minimum put their imprimatur on it. If they
> don't want to distribute it, they should at least push it as the
> "recommended" package. Obviously, I think it would be better if they
> packaged it with their compilers, complete with precompiled object code, but
> that could raise some support issues.
> 
> MDC
> --
> ======================================================================
> Marin David Condic

Make it available in GNAT/GCC form, and I think the vendors may just
end up doing what you asked for. After all, competition is the greatest
motivator ;-)

-- 
Warren W. Gay VE3WWG
http://home.cogeco.ca/~ve3wwg