VOUnits update: Empty/missing units

[Norman Gray]

Mark, hello.

On 16 Dec 2021, at 16:40, Mark Taylor wrote:

> there appear to be two versions of this email that went to the list;
> I'm following up the more recent one.

Subscription confusion, mail queues, email, ... bah.  Yes, the second 
email is the better one.

>> So, to be clear, I think my proposal is:
>>
>>   * Have the VOUnits 'explicitly dimensionless' marker become "1" 
>> rather than "" (on the grounds that the latter is too easily 
>> confusable with NULL/don't know), for the benefit of those metadata 
>> authors who wish to explicitly mark this.
>>
>>   * Adjust the grammars (only VOUnits, or all of them?) to remove the 
>> two-step parse, by permitting parse(<dimensionless marker>) to 
>> produce an appropriate valid result.
>
> I'd be OK with that.  I think it might be slightly surprising
> for people (who haven't read VOUnits 1.1) to come across "1" in the
> units metadata, but I doubt if it's really going to confuse or
> mislead anybody.

Good, and yes, I think this passes a 'I know what you mean' test.

I've confirmed that this change does work in the Unity library, and I've 
added a GitHub issue [1] to note that, if the current rough consensus 
holds up, this should be moved into the VOUnits document.

I suggest that this issues list might be a useful way of logging and 
tracking any remaining (dotting i's and crossing t's) issues in the 1.0 
document, on its way to 1.1.

[1] https://github.com/ivoa-std/VOUnits/issues

>
>> Finally, and as Markus said in the first message in this thread, is 
>> there any case for adding a third possibility: dimensionless / 
>> unknown / not-a-quantity (eg a name)?  That would be very easy to do 
>> in this revision of the document, and would mean that 'the units 
>> field must be non-NULL' would become a reasonable validation 
>> requirement.  But this might be too much detail to hope metadata 
>> authors will supply.
>
> I would prefer not.  Whatever the VOUnits document says, I'm pretty 
> sure
> that people will omit an explicit value for the units in many cases
> where no unit is suitable, and consider themselves justified in doing 
> so.
> Making that an error is just going to have the effect of keeping the
> validators busy while (most) real unit consumers take no-unit-string
> to mean something sensible.

Fair enough.  We don't have to take over the world here!

Best wishes,

Norman


-- 
Norman Gray  :  https://nxg.me.uk
SUPA School of Physics and Astronomy, University of Glasgow, UK