TEAM-ADA Archives

Team Ada: Ada Programming Language Advocacy

TEAM-ADA@LISTSERV.ACM.ORG

Options: Use Classic View

Use Proportional Font
Show HTML Part by Default
Show All Mail Headers

Topic: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Stanley Allen <[log in to unmask]>
Fri, 24 Apr 1998 03:05:05 -0500
text/plain (151 lines)
Chris Sparks wrote:
>
> I need some help in getting some information on the projects that have
> used GNAT for real-time embedded applications.  My lead is concerned
> that GNAT will not be suitable, however, I need proof to dispel his
> concerns.  The situation we are coming across here at work is that we
> are going to use another compiler vendor.  We currently have a
> dependence on VADS.  My lead wants as minimal an impact, however, I
> don't see this happening.  I suggested using some of Ada 95's new
> library support for tasking and he felt that it was not stable enough
> to use.  Any help is always appreciated.
>

Why do you have a desire to use GNAT for this project
rather than the other vendor?  If the other vendor
is not providing Ada 95, then your job is to present
the advantages of moving to Ada 95 as a motivator to
the use of GNAT.  If the other vendor is providing
Ada 95, then you need to know which is the best compiler
for your needs; if you believe this is GNAT, then you
have to present these reasons to your lead.  If it's a
choice between GNAT and a different Ada 95 product and
you want your project to use GNAT, you should re-direct
your question to the GNAT chat list -- I'm sure the
other vendors on the Team-Ada list are not going to
present reasons why GNAT should be chosen for your
project!

If your goal is to get your Ada 83 code to work under
an Ada 95 compiler (later we'll talk about using the
Ada 95 features, but first things first), then it
is up to you to demonstrate that this will cause
no major impact to the project.  Project management
is generally horrified by the prospect of changing
out any of the substrate technology on the project
(the OS, the compiler, the database vendor), even
if it's just going from one Ada 83 compiler to
another.  And depending on the project, this
attitude is quite justifiable.  If your project code
is extremely dependent on VADS-specific features and
implementation quirks, then you've got a tough
row to hoe.  One project in my company tried to
migrate from VADS to GNAT and found it nearly
impossible to do so; another project in my company
doing the same thing had few problems.  The first
project was doing a *lot* of VADS-specific stuff
(also, they were using one of the earlier versions
of GNAT which was not mature enough for the job);
the second had consciously limited dependency on
VADS features.  I'm sure you'll hear the same kinds
of reports (both successes and failures) from people
who've moved from any Ada 83 compiler X to any Ada 95
compiler Y.

My understanding is that Ada 95's designers had this
goal for upward compatibility with Ada 83: that
moving from an Ada 83 compiler to an Ada 95
compiler should be about the same effort as moving
from one Ada 83 compiler to another.  This has
proved true in a number of cases I know of
personally.

Note that all of the above discussion relates to
the situation in which your main goal is to move
your Ada 83 code to an Ada 95 compiler without
making any design changes.  This is absolutely
the first step; if you can't do this with small
impact, you will not persuade your management.

I suggest a pilot project to try compiling your
current Ada 83 code under an Ada 95 compiler.
The goal of this would be to come up with a set
of "lessons learned" notes and a go/no-go
recommendation.

In order to convince your lead that Ada 95 features
should be used, you are better off (1) doing this in
a very project-specific way and (2) recommending only
incremental design changes to make use of specific
Ada 95 features.

Project-specific means that you must take a close
look at your project and take detailed notes about
the Ada 95 features that can supply advantages to the
project.  General arguments about the superior nature
of Ada 95 are not enough.  If there is a weakness in
the current system that can be improved by the use of
an Ada 95 feature, note that.  If there is some external
interface that can be used to good effect in your
system, and that interface is in Ada 95 (for instance
CORBA), then that can be noted.  If portability is an
issue in your project and there is some part of your
system that is currently non-portable because Ada 83
did not provide a portable way to get the job done, it
is possible that Ada 95 has provided a portable way
to do that job.  (When the Ada 95 guys where I work
put together a list the ways in which Ada 95 could
help our project, this aspect was a big factor.  A
great many things that used to be compiler-specific
in Ada 83 are now standard in Ada 95).  If your
project is very performance-sensitive, then note
any Ada 95 features which might improve performance
(like protected objects to replace some tasks).
Our group made a six-page discussion of Ada 95
features that might be of benefit to our project;
the entire discussion was tied to project goals.

It is very important to recommend only incremental
design changes in the early stages of using Ada 95.
There's no reason to throw out all of your current
Ada 83-based design which, even if your project is
only medium-sized, represents a large invest by your
company.  The prospect of redesigning a current system
(at any time, not just when a new technology like Ada 95
is available) is fun for engineers but a nightmare for
project managers.  The engineers see redesign as a great
opportunity to start with a blank slate and "do it right
this time", as well as an opportunity to use fun new
features of the underlying technology (be it the OS or
the implementation language).  Project managers are
generally concerned with the schedule and the budget;
a vigorous redesign is a good way to blow both.

So to start off on the right foot, you must not talk
of using Ada 95 to make major changes to the old design.
This might be your eventual goal.  It may or may not
ever happen.  But it will never happen if you don't get
past the first step, which is to get Ada 95's foot in
the door.  You can't do that if people are convinced that
this will cause a massive re-write of everything done so
far.

Have a list of design changes you would like to see
implemented, describe how these will be beneficial to
the project, and draw up a schedule for prototyping
these changes to determine the cost/benefits of each,
and another schedule which shows how these changes can
ever-so-gradually be introduced into the system.

New projects in Ada 95 do not need so tentative
an approach.  Go ahead, stretch those compilers!
I've noticed that the current crop of Ada 95
compilers is much better than last season's,
especially in those areas that did not see much
use in the early days (like tagged types and
the new generics features), but which now are
starting to be used in new projects.

Stanley Allen
mailto:[log in to unmask]

ATOM RSS1 RSS2