Re: Status of VMS , down to 12 test failures.

[John E. Malmberg]

On 12/3/2014 9:01 AM, Paul Smith wrote:
> On Sun, 2014-11-30 at 19:28 -0600, John E. Malmberg wrote:
> > This is with the bs-nl patch submitted on November 27, 2014 applied
> > and the features/archives patch submitted October 21, 2014 applied.
>
> Hi John; thanks for your work here!  I'll have some time later this
> week/this weekend to review and push these changes.  Sorry for the
> delay...

Delay is not a problem.

Next planned steps:

* Add posix "exit" simulation when the gnv$make_unix_sim is active.
  VMS exit:
      32 bit number.
      Odd numbers success, even numbers various classes of failures.
  Posix exit simulation:
      status < 255 transparently encoded to 32 bits
      status above 255 assumed already in VMS format.

* Review the tests that have not been checked in to see what VMS
  changes can now be backed out.

* Find out why the other 11 tests are not seeing the correct environment 
variables in their test results.

Once all the tests are passing the next planned steps are:

* Change case sensitive mode to be a run-time option based on the VMS
  process setting instead of a compile time option.  Have to figure
  out how to get the verification test for that into the test harness

* Change compiler flags to be the same as with GNU Awk to fully
  support dynamic extension loading.
     /names=(as-is, shorten)
     /float=(ieee) (except VAX)

* Rework the config.h generation to turn on all the options on VMS
  that should be on, but are not, and then make sure that everything
  works as expected.  This will turn on dynamic extensions and
  symbolic link support at a minimum.

Note that on GNU awk, the config.h.in is checked in to git, and VMS now 
just reads that to build the config.h with a set of VMS specific command 
procedures instead of maintaining a VMS specific template.

Regards,
-John