(j3.2006) SC 36 NP (JTC 1 N 8563) and SC 22 NP (JTC 1 N 8567) - Concerns/Objections Due by May 16, 2007

Van Snyder van.snyder
Wed May 2 16:27:54 EDT 2007


While I applaud the notion of modules in C++, N8567 seems to be a bit on
the half-baked side.

The C++ committee should take a close look at how Mesa, Modula-2, Ada-95
and Fortran 2003 augmented by ISO/IEC TR 19767 handle modules.

Especially important is the distinction between definition and
implementation in Modula-2 terminology, or specification and body in Ada
terminology, or module and submodule in Fortran terminology, all
stemming from Mesa.  Ada divides the N8567 concept of "module partition"
into private child units and public child units.  Modula-2 has no child
units.  Fortran (as of TR 19767) has only private child units.

When Niklaus Wirth developed Oberon, he abandoned the distinction
between definition and implementation in favor of a monolithic module,
as in Fortran 90.  Monolithic modules have proven to be suboptimal for
large software projects, an observation to which Wirth now agrees.

Best regards,

The following SC 36 and SC 22 NPs are currently undergoing SC ballot and
JTC 1 National Body review:


      * JTC 1 N 8563 ? SC 36 Proposed NP on Nomadicity and Mobile
      * JTC 1 N 8567 - SC 22 Proposed New Work Item - Programming
        language C++ - Modules


Concerns with/objections to these NPs should be sent to me
<jgarner at itic.org> (202.626.5737) no later than May 16, 2007 for
consideration by the INCITS Executive Board as possible USNB comments on
the JTC 1-level reviews of these NPs.  Please note that the appropriate
US TAGs will submit the actual USNB votes on these SC-level NP ballots.



JTC 1 Coordination Register
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://j3-fortran.org/pipermail/j3/attachments/20070502/2d803c70/attachment.html 

More information about the J3 mailing list