TJ wrote:
> Over the last couple of days I've been working on the ".sdf" issue.
> It should now be fixed so you can say
>
> stats $KAPPA_DIR/comwest.sdf
> hdstrace $KAPPA_DIR/comwest.sdf.DATA_ARRAY
>
> if you are so inclined.
What about the following?
hdstrace $KAPPA_DIR/comwest.SDF.DATA_ARRAY
I would imagine that this would look for a .SDF component, given
DAT__FLEXT.
> The starting point was to allow for the .sdf to be present in filenames.
> Because of code duplication this needed fixing in at least NDF, NDG and
> SUBPAR (DAT_ASSOC). Since I did not want to change 3 pieces of near identical
> logic I refactored....
Excellent. I hope that there's more time for this. I certainly concur
with David's point [in another discussion] about avoiding historical
binding of code associated with particular author's packages. The
general KAPPA and CCDPACK routines ought to be organised by function
object, e.g. image processing, statistics, graphics; or object like
FITS, graphics database, AST Frame, NDF/HDS objects. The danger is
having too many libraries offsetting the ease of location of
functionality. As you've seen some will go into existing libraries.
Also we'd reduce the code base. For instance, swathes of ESP would go.
> * Moved a large amount of code that was duplicated in NDG and NDF into
> HDS. It was all very useful code for analysing an HDS component path and
> it was so useful I put it in HDS:
>
> HDS_FIND - find a locator to a component given full path to the
> component
Brilliant.
> DAT_CUT - Like DAT_SLICE but takes a string
Again long overdue.
> CHR_FPARX - this routine (aka NDG1_FPARX and NDF1_FPARX) simply
> finds matched parentheses in strings. So I moved it to CHR.
Is there one for brackets too?
> * Whilst doing the above I was horrified (but not surprised) to learn
> how many libraries and applications were using xxx_LASTO() subroutine.
> NDF, NDG, 3 * KAPLIBS, CCDPACK, PISA and KAPPA to name a few. I moved
> that routine to CHR and fixed up all the libraries to use it.
> [and for Malcolm: Tests were added and the CHR SUN has been updated]
Thanks. I just worry that in the enthusiasm, the boring documentation
is left behind. And yes I'll be updating SUN/95 for CLINPLOT once the
COLLAPSE modifications are completed.
Malcolm
|