[Imgcif-l] Better organization for a named array reference?

Joe Krahn krahn at niehs.nih.gov
Mon Mar 19 16:18:11 GMT 2007


In looking over a proposed 3D 'map' format, I found that a data
reference required two references to name a block of data, an array_id
(which really should be called structure_id) and a binary_id. This is
due to the way array_data and array_structure categories are defined. It
seems that this is probably a 'legacy' issue from the original simple
binary block format, before binary strings. It may also be that the
original design was oriented to single-image data blocks. In any case,
the current organization seems rather convoluted in terms of how to
select an array object.

(I commented on this to Herbert previously, so some people may have
already considered this.)

I propose a simple addition to allow for a single reference to select a
binary array by name. To allow for this, a single new category could be
added, perhaps called just 'array_name'. This category would list all
valid structure+binary pairs, and give each a name. Then references to
array_id+binary_id could be replaced by just array_name. Ideally,
'array_id' should be replaced with 'structure_id' everywhere to be more
accurate.

Alternatively, if people want to keep structure.id as the primary
array-name reference, then the structure_id category should include a
binary_id reference for each array_structure member.

I am assuming that the binary-string form is still new enough that some
organizational changes are not so bad. A few changes now while it's
still new is better than being stuck with some 'legacy' issues for years
to come.

Joe Krahn


More information about the imgcif-l mailing list