🚀 go-pugleaf

Submitted-by: stephe@mks.com (Stephen Walli)


>In article 1halvbINN9kd@ftp.UU.NET, stephe@usenix.org (Stephen R. Walli) writes:
>Actually, producing a language independent specification plus language
>bindings is already an existing practice for API standards.  Both the
>ISO/ANSI GKS and PHIGS standards have this form.  The only aspect of the
>POSIX effort that cuts new ground is the use of LI specifications to
>approximate existing APIs.

When IEEE POSIX began the LIS efforts, the ISO GKS model was investigated
and found to be wanting. Hopefully without sounding trite, the GKS LIS 
proved you could write Fortran programs in any language.  POSIX is trying 
to come up with a model that will take into account different language 
models, the three it has to take into account immediately being: C, Ada, F77.
A lot of effort has gone into this. The IEEE POSIX person responsible 
for the TCOS LIS Guidelines *is* the liaison to ISO WG11 (Language Binding
Techniques) and we have already overreached the WG11 work in a number of 
areas. Operating system APIs are apples. Graphic objects are oranges. 

The requirements placed on POSIX were different from the start in that
there were anywhere from 6-12 Fortran programmers, and 20-30 Ada
programmers looking over our shoulders at any one point in time.  These
were people who were active in their own working groups, so only had
time to provide critical comment over the wall. In separate
conversations, I have heard a number of Ada programmers complain
bitterly about the GKS Ada binding.


------------------------------------------------------------------------------
Stephen R. Walli,       stephe@mks.com | Standards are commercial and  
Mortice Kern Systems Inc.              | international politics dressed up  
phone: +1 (519) 884-2251               | as technical arguments. 


Volume-Number: Volume 29, Number 92