ClearCase and Ada -- Config. Management

ClearCase and Ada -- Config. Management

[Daniel Schwepker]

Currently we are using ClearCase for Configuration Management on an SGI
platform. This works well for C and Fortran code. Ada is a different
matter. ClearCase has its own intelligent make utility, Clearmake, that
keeps track of timestamps and dependencies on its own. Rational Ada
also keeps track of its own timestamps and dependencies, as it has to.
When I put Ada under ClearCase control, Clearmake always will rebuild
every Ada compilation unit every time. But if I only use a.make, I do
not get the advantages of ClearCase sharing object files between users,
or users being able to work in separate "views". Is anybody currently
developing Ada under ClearCase? What contortions have you had to go 
through?

Daniel

Re: ClearCase and Ada -- Config. Management

[Stef Van Vlierberghe]

In article <36hjek$18k@hacgate2.hac.com> dschwepk@samson.hac.com (Daniel Schwepker) writes:

   Currently we are using ClearCase for Configuration Management on an SGI
   platform. This works well for C and Fortran code. Ada is a different
   matter. ClearCase has its own intelligent make utility, Clearmake, that
   keeps track of timestamps and dependencies on its own. Rational Ada
   also keeps track of its own timestamps and dependencies, as it has to.
   When I put Ada under ClearCase control, Clearmake always will rebuild
   every Ada compilation unit every time. But if I only use a.make, I do
   not get the advantages of ClearCase sharing object files between users,
   or users being able to work in separate "views". Is anybody currently
   developing Ada under ClearCase? What contortions have you had to go 
   through?

   Daniel

We have done a test using ClearCase with Alsys Ada a while ago. In a
nutshell, we used ClearCase to manage a network of subsystem baselines,
holding the sources in ClearCase and building the associated libraries
outside ClearCase, using the Ada make utilities. In this model all
development was done in per-developer personal libraries that depend
only on a set of frozen libraries.

The important aspect of Ada builds is that the compiler vendor is normally
best placed to deduce dependencies from sources, because it is the
compiler that knows about generic stubs/inlines, accepted pragma INLINE
advise, automatic inlining, which is quite a bit more than just getting
to the "with", body-to-spec, subunit-to-parent dependencies. For Rational
Ada this goes even a lot further. As far as I know, Clearmake is C-based
tool that is based on looking for dependencies based on #define cpp
directives, I don't quite see how this would be useful in an Ada environment.

--
Stef VAN VLIERBERGHE            Eurocontrol - Central Flow Management Unit
stef@cfmu.eurocontrol.be        Avenue des Arts 19H
Tel: +32 2 729 33 42            B-1040 BRUSSELS
Fax: +32 2 729 32 16            Belgium

Re: ClearCase and Ada -- Config. Management

[Robert Dewar]

There is an active project underway at SGI to interface Clearcase to 
GNAT, and this combination will, as I understand it, be the basis of
the SGI GNAT/based Ada product. Conctact Wes Embry at SGI for further
details.

Re: ClearCase and Ada -- Config. Management

[Tucker Taft]

ClearCase should work relatively well with Ada compilers that
have a source-based library, such as GNAT.  However, if the compiler 
maintains an intermediate database, everything will probably end up being 
dependent on central parts of the database that are continually changing
(e.g. some kind of index).

Perhaps ClearCase has a way to delete dependencies, to avoid this problem
of volatility in a central database.  You might want to contact Atria.
It seems that many non-Ada tools are beginning to have some kind of
central database, and at least the index files of such a database
would create similar problems, so I doubt if this is an Ada-only problem.

-Tucker Taft   stt@inmet.com
Intermetrics, Inc.