Thanks Brian. Brad reports that his ORAC-DR test now completes in 64-bit
mode!
When you've done the DAT_PUT/GET change I think it will be time to move
this all on to the main CVS HEAD. [the default fallback mode seems pretty
stable anyway]. I committed a couple of patches that were preventing
compilation with gcc4.
A third patch is left uncommitted since you may have an opinion: The
dat1.h prototype for dat1_get_off differs from the actual declaration for
dat1_get_off by the use of "static". GCC4 refuses to compile. Can you sort
that out please?
As an HDS aside, I'm getting this error from an application built using
gcc4/g95
!! O111.MORE.SCUBA.DEM_PNTR: Unknown error: 149324595
! SURF: bad status on exit from REDUCE_SWITCH
OBEYW unexpected status returned from task "surf_mon11325", action -
"REDUCE_SWITCH"
ADAMERR %CMP, Data currently mapped
but *only* when running in A-task mode. The command line is fine. This is
from a CMP_MAPV. HDS_SHOW does not indicate that .MORE.SCUBA.DEM_PNTR has
been mapped previously (only the DEM_PNTR in the input file STRUCT.MORE
not the output O111.MORE.):
O111.MORE.SCUBA, group=NDF_DCB
O111.HISTORY.RECORDS, group=NDF_DCB
O111.HISTORY, group=NDF_DCB
O111.MORE, group=NDF_DCB
O111.DATA_ARRAY, group=ARY_DCB
O111.DATA_ARRAY, group=ARY_DCB
O111, group=NDF_DCB
SURF_MON.REDUCE_SWITCH.SWITCH, group=P0159
SURF_MON.REDUCE_SWITCH.SPIKE_LEVEL, group=P0158
STRUCT.MORE.SCUBA.DEM_PNTR, group=NDF_DCB
STRUCT.MORE.SCUBA, group=NDF_DCB
STRUCT.DATA_ARRAY, group=ARY_DCB
SURF_MON.REDUCE_SWITCH.USE_CALIBRATOR, group=P0157
STRUCT.MORE.FITS, group=NDF_DCB
STRUCT.MORE, group=NDF_DCB
STRUCT, group=IN
STRUCT, group=
STRUCT.DATA_ARRAY, group=ARY_DCB
STRUCT.DATA_ARRAY, group=ARY_DCB
STRUCT, group=NDF_DCB
SURF_MON.REDUCE_SWITCH.MSG_FILTER, group=P0156
SURF_MON.REDUCE_SWITCH.ADAM_DYNDEF, group=PROGRAM
SURF_MON.REDUCE_SWITCH, group=PROGRAM
SURF_MON, group=
Any thoughts on where I should look to debug this? It's present in old and
new HDS in 64- and 32-bit mode.
Tim
On Tue, 16 Aug 2005, McIlwrath, BK (Brian) wrote:
> Hi Tim,
>
> A fix for hdsCopy is now committed back into CVS. I am thinking about
> the _CHAR variable problem in DAT_PUT/GET as the next stage.
>
> Brian
>
> Tim Jenness wrote on 15 August 2005 19:03:
>
>> sorry. Don't know how that happened. Try this.
>>
>> On Mon, 15 Aug 2005, Brian McIlwrath wrote:
>>
>>> On Tue, 9 Aug 2005, Tim Jenness wrote:
>>>
>>>> problems with basic copying. The attached fortran code
>>>> demonstrates the problem. It fails when HDS_64BIT is set with:
>>>>
>
> ----------------------------------------------------------------------
> Brian McIlwrath, e-mail: [log in to unmask]
> DCC/Starlink, Tel: +44 (0)1235 446254
> Space Science & Technology Department, Fax: +44 (0)1235 446362
> Rutherford Appleton Laboratory, UK.
> ----------------------------------------------------------------------
>
>
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|