Re: Another measure of Ada's rejection by corporate America

[Howard.Gilbert@yale.edu]

In <3iepqn$6ej@mica.inel.gov>, "Paul H. Whittington" <phw@inel.gov> writes:
>
>An axiom in the marketing game goes something like this:
>
>  In order to break into an existing stable market (C++ etc.)
>  a new product (Ada) must either offer twice the PERCEIVED
>  positive characteristics at the same price, when compared to
>  the existing products in the market, or offer the same
>  PERCEIVED positive characteristics at half the price.
>
The C++ market is hardly stable or established.  Borland's 
class library is incompatible with MFC.  The PC market is 
undergoing a transition from 16 to 32 bit compilers.  The 
relationship between C++ and WIN32 system services is not
clear.  Mostly, people use the C subset of the C++ compilers.

At a larger scale, there is a major uncertainty whether the bulk
of development will use compiled languages like C++ or 
interpreted languages like Smalltalk.  Will objects be internal
like C++ or external (Corba).

By analogy, there is a large installed base of ONC (Sun) RPC 
stuff, and it is largely free or at low cost.  Yet the industry is
moving (slowly) in the direction of DCE even at some cost and
conversion because there is a need for security and better
overall structure.  This runs directly against the previous
axiom, since DCE is not cheaper but has a more intangible benefit 
(security).

>
>    The perception here is that the C++ market can provide, in a
>    timely manner, tools that, at the same time, provide for rapid
>    development of applications, with acceptable quality, and support
>    for the latest enabling technologies (OOP, OLE, MAPI, TAPI etc.)
>
Success is measured in Bindings, not in the base language.  Put 
another way, Visual C++ 2.0 will succeed or fail based on how people
react to the Microsoft Foundation Classes and the development 
environment, not based on C++ itself.  I have no data, but I 
wouldn't be surprised if 99% of the C++ users never build a single 
substantive class of their own, but just use vendor supplied class 
libraries.  In the past, Ada has been limited by a lack of bindings 
to widely used systems.  This despite the fact that 
keyword arguments with defaults makes such bindings much easier to 
develop in Ada than other languages.

If people use base system services (WIN32, OS/2 API) then Ada can 
greatly simplify programming.  If people expect to use massively 
repackaged services through objects (MFC) then it is more confusing. 
If they really are unable to deal with programming at all (VB, 
Visual Age) then the whole issue is moot.

>
>    The fact is Ada development products pale in comparison with
>    their C++ peers, not to mention their Smalltalk and PASCAL
>    peers.  We have to bring Ada development products at least up
>    to par with C++ or there is no hope.  If Ada is really as
>    capable a language as we all like to think it is then we should
>    be able to provide considerably more positive characteristics.

Its the secondary elements that tell.  A lot of Ada compilers 
started out with no debugger at all.  The next level is to have a
dbx level of debugger.  If people expect something along the lines
of the MS Debugger or the IBM IPMD, then that looks cheezy.  Since
the other debuggers exist, and since they clearly support C++, the 
assumption is that the issue is generating the right object format.
For example, if OS/2 GNAT moves over to EMX then it will inherit 
(at least in theory) the ability to be debugged with IPMD.  The 
WIN32 equivalent is a bit muddier since it is not as clear who is 
in charge of that port.

But hope is not completely lost.  Drop Ada code in front of most 
"real world" programmers and they will say, "Oh, some Oracle PL/SQL"
because that language is largely based on Ada statement syntax. 
But there is a lot of Oracle around, and people who would never 
consider "Ada" by name are learning much of the basic syntax 
indirectly without knowing it.

---------------
Howard Gilbert -- Chief Mechanic at PC Lube and Tune
Technical training on PC's, networks, and communications.
Point Netscape or WebExplorer at http://pclt.cis.yale.edu/pclt/default.htm