RE: [Gcl-devel] 2.6.2.....

[Mike Thomas]

Hi Camm.

| Great!  On the standard tests, we no longer have any issues with
| compiler version and/or C optimization level, provided
| -fomit-frame-pointer is removed, right?

As mentioned in an earlier email the release package will use and require
gcc 3.3.1 as used for the tests.

I just built GCL with gcc 3.3.3 and binutils 2.15.90 (20040222) and then
Maxima and got Vadim's error.  I expect gcc 3.4.0 is also not going to work.


==========================================================================
;    - Compiling module "special-functions"
;      - Compiling source file
;        "c:/lang/source/gcl/maxima-2004-04-24/src/specfn.lisp"
Compiling c:/lang/source/gcl/maxima-2004-04-24/src/specfn.lisp.
End of Pass 1.
End of Pass 2.
OPTIMIZE levels: Safety=2, Space=2, Speed=2
Finished compiling binary-gcl/specfn.o.

;      - Loading binary file "binary-gcl/specfn.o"
Loading binary-gcl/specfn.o
Error in CONDITIONS::CLCS-LOAD [or a callee]: Caught fatal error [memory may
be
damaged]

Fast links are on: do (use-fast-links nil) for debugging
Broken at FILLARRAY.  Type :H for Help.
 1 (Continue) Retry loading file "binary-gcl/specfn.o".
 2 (Abort) Return to top level.
dbl:MAXIMA>>
==========================================================================


| Just a note here -- 2.6.2c2 should not only run without crashing, but
| should report no errors.

Is some kind of auto logging of error forms retained?

My understanding is that if a form is printed on the console during the test
then it has failed and that is the only record.  Is that correct?

I have been using the head branch RT as I was under the impression it had
not been backported to stable - I now understand I was mistaken.

My repeat test with the stable branch RT so far has done 7300 (200 3)
iterations without error.


| > I copied the 2.6.2 source into a modified Axiom source tree but
| the build
| > dies with the error shown at the very bottom of this email.  This I will
| > also leave to the experts.
| >
|
| :-(.  OK, this is (so far the only, to my knowledge) test 2.6.2 will
| fail on windows.  We'll get to this in 2.7.x.

Agreed re 2.7.0.

I believe that I may not have done the test I thought I had done as I keep
getting output from the Axiom build consistent with a 2.7.0 version of
"o/unexnt.c"  - either I've copied the wrong directory, the file is being
patched durin the Axiom build, or I've got something else wrong.  I'll
investigate further as I would like to clearly delineate the boundaries of
2.6.2

Cheers

Mike Thomas