Re: How to get a »Conventional Ada Library« (Was: Ideas for Ada 200X)

[Marin David Condic <nobody@noplace.com>]

Just for the record, the quoted text was mine and not that of Jacob 
Sparre Andersen.

I don't understand what you mean about "Fighting the wrong battle." 
First, I wasn't saying that a library needs to be "Complex" in order to 
succeed. I said it needs acceptance by the vendors in order to succeed. 
There may be those who would care to dispute that point. I think its 
intuitively obvious to even the most casual observer. :-) One may argue 
about how to get there, but I think that is a necessary condition to 
success.

As for the CAIS - I remember discussions and papers about that but don't 
recall the specifics. I remember not understanding what it was trying to 
achieve or how it was supposed to work - that ought to be an indication 
it was doomed. ;-) If you can't understand what it is attempting to 
achieve and how it is going about getting there, its a warning flag.

As for a conventional Ada library getting too complex, let me start 
here: To the extent that containers are part of such a library, I'd 
support a plain-vanilla set of lists and maps with "Ease Of Use" as a 
primary goal. I'd also want to consider a second set of more elaborate 
and powerful containers with "Maximum Utility" as a goal - on which you 
go build whatever else you want to stick into the library. Make sure it 
is well documented.

Second, as for any concerns about "largeness" or "complexity" - lets do 
a counter-proof by existence: Java has a really large library. So does 
MSVC++. Programmers use those libraries *every*day* to build systems. If 
they count as "large" and "complex", then this must not be a handicap. 
If they are not "large" or "complex" then let's go off and build 
something similar to what they have and call that a Conventional Ada 
Library and we would similarly not be "large" and "complex".

One advantage to making sure we have a "Conventional" Ada library is 
that it makes it much easier for the academic types to write books about 
how to use it. I'd think it is really important to have good 
documentation with the library, but secondary books that discuss the 
"How Tos" of program development with the library would be just as 
important. I've got books on the "How Tos" of using MSVC++ and the MFC. 
I presume they exist for Java. Why not for Ada?

MDC



Robert I. Eachus wrote:
> Jacob Sparre Andersen wrote:
> 
>>> Well, I won't say it *can't* work - just that it has been tried in a 
>>> few different ways and guess what? We don't have a Conventional Ada 
>>> Library.
>>>
>>> If the thing has zero interest from the vendors it will become Just 
>>> Another Ada Library like the half-dozen or so that are already out 
>>> there.
>>
> 
> You are fighting the wrong battle here.  Back in 1984 there was a 
> session at SIGAda to discuss CAIS a government developed APSE 
> (development environment) for Ada.  We had a very lively discussion for 
> about an hour, and I then decided to see if I could cut through the all 
> the cross-currents and come up with a single position statement.
> 
> I started out by saying that I had spend three weeks studing the CAIS 
> documentation, and I still wasn't sure I knew enough to use it as a 
> development platform.  So I asked those people present who had use the 
> CAIS prototype how much study by programmers was required to use it.  No 
> answer.  Does anyone feel that a month per programmer is an 
> overstatement.  This time I got a lot of heads shaking no.  Is there 
> anyone here who has spent less time?  No answer.
> 
> Okay, I then asked people present who had developed fairly significant 
> Ada projects that had been ported to a different environment, how long 
> the ports had taken.  "About three weeks."  What if you had to do more 
> ports?  "Much easier."  Now I asked the one group that had ported CAIS 
> to a new environment how long that had taken.  "About three months."
> 
> That was basically the end of the meeting.  We wrote up a statement 
> saying that the complexity of the CAIS environment was such that we 
> didn't feel that using it for development would ever pay off.
> 
> Why bring this up?  I think this is currently where we are with Ada 
> container libraries.  A SIMPLE Ada container library will get used.  But 
> it is much easier for the developers to "improve" the libraries by 
> adding new container types and modes than it is for the potential users 
> to learn how to use the library.  I figure that about five to seven 
> pages of technical documentation--including examples--is what Ada 
> programmers are willing to spend on a container library before walking 
> away and reinventing the wheel.  This doesn't say that the library can't 
> have all those bells and whistles.  Just that the standard subset has 
> got to be that easy to learn.
> 


-- 
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jast.mil/

Send Replies To: m c o n d i c @ a c m . o r g

     "In general the art of government consists in taking as
     much money as possible from one class of citizens to give
     to the other."

         --  Voltaire
======================================================================