Example VOTable for v1.6

[mbtaylor]

This PR is to provide an example VOTable with non-ASCII content in char and unicodeChar FIELDs and PARAMs, which may be useful for implementors who want to test VOTable 1.6 I/O code.

I think it is useful to make test data like this available, but since this example table is not included in the VOTable specification or otherwise part of the VOTable document build system, I don't know whether this repository is the right place for it. I'm happy to make the test table available elsewhere instead if somebody can think of a good place.

Opinions welcome.

[msdemlei]

I think such an example file is an excellent idea, and I think this repo is a good place for it. Actually, I'd say this is a good use case for an ivoatex \auxiliaryurl, where we'd have a VOTable with O(2) subtables exercising the main tricky features (arrays and NULLs, too), linked to in a footnote from the opening paragraphs of section 5.

Why ~two tables? I'd like to see both TABLEDATA and (in this case more interesting) BINARY2, both of which would have the same content.

I give you that's a steep-ish agenda, but I'd have loved to have such a thing when developing my VOTable code. I'd pledge four hours of work on such a thing.

[mbtaylor]

I can easily generate a BINARY2 version of the same thing:

   stilts votcopy in=unicode-example.vot format=binary2

(though you need the pre-release stilts from topcat-full_votable1.6.jar to run that). And stilts can also generate a load of generic (non-VOTable-specific) test data with different datatypes and nulls that you can use to generate a test VOTable, e.g.:

   stilts tpipe in=:test:10,ibsfwm ofmt='votable(format=binary2,version=v15)'

Maybe rather than one test table we'd be better off with a subdirectory where people can add a variety of test tables including interesting edge cases they come across?

[msdemlei]

On Tue, Feb 03, 2026 at 12:42:11PM +0000, Mark Taylor wrote: Maybe rather than one test table we'd be better off with a subdirectory where people can add a variety of test tables including interesting edge cases they come across?

As long as people reading the spec have a chance of spotting the "exercise data" and it ends up downloadable from the doc repo, too (as opposed to only from github), I'm fine with anything.