fntests.m during and after make

[Ben Abbott]

On Dec 3, 2007, at 10:20 PM, Ben Abbott wrote:

>
> On Dec 3, 2007, at 12:07 PM, John W. Eaton wrote:
>
>> On  3-Dec-2007, Ben Abbott wrote:
>>
>> | I encountered a "panic" during "make check" for Octave 2.9.17
>> running
>> | Mac OSX 10.5 (Leopard) on Intel.
>> |
>> | I thought I'd check the remaining test and also try to isolate the
>> | problem. To do so I installed octave and ran "fntests.m" manually
>> from
>> | within octave.
>> |
>> | ... to my surprise, everything passed!
>> |
>> | Can someone explain? ... is this to be expected or is there a bug
>> | lurking about?
>>
>> I think you're going to have to provide more details.
>>
>> If you think you've found a bug in Octave, please send a *complete*
>> report to the bug at octave.org list.  Please read
>> http://www.octave.org/bugs.html to see what to include in your report
>> so that it is useful.
>>
>> jwe
>
> I've located a specific failure, but am not sure it is a bug, or a
> feature of the way I've build octave.
>
> "test datevec" produces a kernel panic. After some testing I found
> that any date string with either ":" or "/" give that result. Thus,
> each of the following ...
>
> datevec("07-Sep-2000 15:38:09")
> datevec("07-Sep-2000")
> datevec("09/07/00")
> datevec("09/13")
> datevec("15:38:09")
> datevec("3:38:09 PM")
> datevec("15:38")
> datevec("03:38 PM")
> datevec("03/13/1962")
>
> ... produce
>
> ---------------------------------
> panic: Bus error -- stopping myself...
> attempting to save variables to `octave-core'...
> save to `octave-core' complete
> Bus error
> ---------------------------------
>
> While the following work as expected
>
> datevec("07-Sep-2000 15:38:09")
> datevec("07-Sep-2000")
> all(datenum(datevec([-1e4:1e4]))==[-1e4:1e4]'))
> t = linspace (-2e5, 2e5, 10993);
> all (abs (datenum (datevec (t)) - t') < 1e-5));
>
> I've built octave on an intel Mac using Fink's package management. The
> config parameters and build script are below
>
> ---------------------------------
> ConfigureParams: F77=%p/bin/g95 --libexecdir='${prefix}/lib'
> CompileScript: <<
> #!/bin/sh -ev
> gcclib=`%p/bin/g95 --print-lib`
> FLIBDIR=`dirname $gcclib`
> export FLIBS="-L${FLIBDIR} -lf95"
> export FFLAGS='-O2'
> export CFLAGS='-O2'
> export CXXFLAGS='-O2'
> ./configure %c
> make
> make check
> ---------------------------------
>
> I'm hoping someone will point out that I've made an error. If not I'll
> file a bug report.
>
> Ben
>

oppps, the failures should have been

datevec("09/07/00")
datevec("09/13")
datevec("15:38:09")
datevec("3:38:09 PM")
datevec("15:38")
datevec("03:38 PM")
datevec("03/13/1962")

Ben