[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

alpha-7 release notes.

The change to xpp/warrayx.cxx was sufficient to make the xpp regression
tests run, and the crashing of the calc example has now been characterized.
Therefore QA the recall is over.  (Sorry it took so long, but the bug that
ate the calculator was way off in left field.)

If you already grabbed the files, you should get a new copy of:


Status of the various subdirectories:

  fm, platform/macintosh:

   - (Empty.)

  febe, platform/sunos, server:

   - Untested.


   - No bugs detected.


   - The regression tests have not been updated, so all of them still
     claim to fail.


   - The makefile does not yet include dependencies for gmon-.o
     (Easiest immediate workaround:  "cd xpp; rm gmon-.o;
     make gmon-.o" before other makes.)

   - Previously allocated objects may be overwritten during later
     object allocation on the EPISODE heap.  This breaks server/client
     operation.  The EPISODE heap is apparently used only by
     SocketTransporters, so only server/client operation should be

Some notes on the "release notes":

They are intended to keep others informed of bugs I've found while
grabbing and testing the new release, so the maintainers of the
modules can fix or upgrade, and the other users don't waste time
tracking down the ones I've already found.  They should not be
interpreted as a criticism of the authors or the heroic efforts
of our integrator (who deserves an ovation).

You will note that I only ran the tests in the directories that
contain the files I'm using (which I had to do anyhow).  I may do
more on releases that occur when the boxes on my line of the PERT
are running a cooler color.

The elapsed time of this testing (three days) was an artifiact of
the subtility of the bug I had to trace to characterize the calc
problem compounded by a number of grand-compiles that occurred
because I was low on disk space and/or otherwise klutzed things.
An additional factor was the need to use unfamiliar tools to test
other people's code, rather than certify my own.  The learning
curve on the second and third factors should be steep.