Experimental Physics and
Industrial Control System

Eric Norum <[email protected]> · Wed, 17 Sep 2008 10:29:52 -0500

I don't know about CEXP -- Ernest will have to check with Till on this.

I did try running the tests on a couple of platforms here. Looks like

RTEMS handles the C++ destructors properly.

==================================================================

uC5282 -- has a few flaws in the corners of the floating point support

routines:

not ok 259 - -Inf * NaN
# Expected result is nan, actually got inf
        Literal: inf
        UNARY_NEG
        Literal: nan
        MULT
ok 260 - NaN * 0.0
not ok 261 - NaN * Inf

# Expected result is -inf, actually got nan <<<<<< Interesting --

the CALC is right and the C code is wrong!

        Literal: nan
        Literal: inf
        MULT
.
.
.
not ok 311 - -Inf / 1.0
# Expected result is -inf, actually got inf
        Literal: inf
        UNARY_NEG
        Literal: 1
        DIV
.
.
.
***** epicsAlgorithm *****
1..22
ok  1 - epicsMin(f1, f2)
ok  2 - epicsMin(f1, -Inf)
not ok  3 - epicsMin(f1, NaN)
ok  4 - epicsMin(f1, Inf)
.
.
.
    EPICS Test Harness Results
    ==========================

Failing Program           Tests  Faults
---------------------------------------
epicsAlgorithm               22       4
epicsCalcTest               533       3

Failed 2/20 test programs. 7/1587 subtests failed.
Programs=20, Tests=1587, 597 wallclock secs

<<<<<< Interesting -- the CALC is right and the C code is wrong!

====================================================================
mvme3100 -- looks good
    EPICS Test Harness Results
    ==========================

All tests successful.
Programs=20, Tests=1587, 458 wallclock secs

On Sep 17, 2008, at 9:54 AM, Andrew Johnson wrote:

Hi Ernest,

On Wednesday 17 September 2008 00:01:58 Ernest L. Williams Jr. wrote:

Target:  MVME3100
OS: vxWorks 6.6
EPICS: base-R3-14-10-pre1

Looks like, I failed on an exception.

The libCom tests are meant to be a standalone program run in a

system without

an IOC running. Please boot vxWorks with no startup script, load

*only* the

vxTestHarness.munch file and then execute epicsRunLibComTests — if

everything

passes you should only need to post the overall summary message

printed at

the end (at which point you'll have to Ctrl-C to get a shell back).

RTEMS is

going to be similar using rtemsTestHarness.elf, although I'm not

sure if we

build it properly for the CEXP configuration — Eric?

- Andrew
--
Talk is cheap. Show me the code. -- Linus Torvalds

--
Eric Norum <[email protected]>
Advanced Photon Source
Argonne National Laboratory
(630) 252-4793

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System