Re: Ada and Mac (Was: New version of AppletMagic)

[mfeldman@seas.gwu.edu (Michael Feldman)]

In article <JSA.96Oct17172950@alexandria>,
Jon S Anthony <jsa@alexandria> wrote:

>Well, OK, you have me.  I appears I am just plain wrong.  I really
>don't know the contract specifics and thought it was simply requiring
>a readily accessible and cheap "Academic Ada Compiler".

There was a good bit more than that in the RFP. Full disclosure:
we (GW, with yours truly as team leader) submitted a proposal
against this RFP, so I know its contents fairly well.:-)

>OTOH, if the mod was OK'd by the customer, how is this still not
>"fine" (well, other than it goofs up the Mac situation).

Because it goofs up the Mac situation. 

At the time of the contract award, we got very good reviews on the
technical merits, but because we proposed GNAT as the core compiler,
we were ruled out essentially because at that time (mid-1994) there
was no way to _promise_ either validation or ongoing _commercial_
backing for the products. This may seem bizarre, because one of GNAT's 
key purposes was to support education; we would have provided the
free or nearly Intel and Mac compilers-with-IDE's that were
contemplated in the RFP. But it wasn't to be.

I can justify the government's position somewhat - they wanted to see 
their (ummm, the taxpayers') $1.5 million go to kick-start a _commercial_ 
cheap-compiler industry (at the time, GNAT was not seen as commercial). 

To the extent that funding helped to complete the AdaMagic front end and 
provide the basis for what is now Thomson ObjectAda (which uses that front 
end) one of the goals was met. IMHO, the situation would be far better if
the sponsor hadn't let the contractor walk away from the Mac. 

It is, of course, not too unusual for a bidder to promise everything in 
an RFP, then - after the contract is won and it's too late to re-bid it - 
negotiate for a mod. You are right - if the customer accepts the mod,
well, that's life. But _this_ mod undermined one of the goals of the
original project, namely to fill the Mac gap.

Luckily, GNAT's commercial maturity, coupled with a change of 
administration at AJPO, made it possible for us to apply successfully
for funding (under ATIP-P) to do the current GNAT-Mac project. But 
the funding was _significantly_ less, required significantly higher cost
-sharing, and required completion on a much tighter schedule (less than a
year end-to-end). 

So we did the best our resources allowed, namely, building on
an existing Mac-based GCC and using the existing GNAT command-line
tools instead of building yet another GUI IDE. We've met the contractual
requirements (without any mods!) and are filing off the rough edges for 
the commercial product.

And, if we're all lucky, the resulting CodeBuilder CD will be
popular and break the Mac/Ada deadlock.

Mike Feldman