DALI 1.1 comments

[Paul Harrison]

> On 2016-05 -10, at 15:37, Arnold Rots <arots at cfa.harvard.edu> wrote:
> 
> On the ISO-8601 question of the 'Z':
> 
> FITS forbids the use of Z, for good reasons.

Then FITS should not even be mentioning ISO-8601 as not having the Z in ISO-8601 means that the date time string should be interpreted as a local time.

The issue is that there are many libraries in many computer languages that implement ISO-8601 and will make the above error when used on ISO-8601 formatted strings without the Z - for interoperability with software systems that have not specifically been designed with VO compliance, then it would be better if something that looks like an ISO-8601 timestamp is really an ISO-8601 timestamp - it can then be simply cut and pasted. It seems to me that by not having the Z the IVOA is trying to swim against the tide.


> 'Z' is really intended to provide a shorthand timezone designation,
> distinguishing UTC from MET, EDT, PDT, etc.
> That makes sense in a context where time is measured on the globe,
> so, I guess, it makes sense for OAI.
> In astronomy we have no use for timezones, but use time stamps
> tied to specific time scales, one of which is UTC.
> Therefore we do not need the 'Z'.
> We do need to be explicit in our specification of time scale and
> should attach that label to all our time stamps, implicitly or
> explicitly.
> Clearly, writing "2016-05-10T14:33:03Z (TDB)" would be insane.
> If, therefore, the 'Z' would either be allowed or required for all
> UTC time stamps, we would end up with two independent
> time scale designations, which is unwise, silly, and dangerous.
> Besides, the use of the 'Z' is currently prohibited by STC 1.33,
> so allowing or prescribing it would require a change in an existing
> standard.

Clearly the focus of ISO-8601 is the conversion of string representations of timestamps onto "civil time" and has nothing to say about timescales - but again this in why I feel that the FITS standards are being duplicitous in citing iso-8601 and then not following it.

I would suggest that to write any times in something like iso-8601 format when you want to express a time in a particular timescale (other than UTC) is the act that should be banned, as it is highly likely that your intentions will be misinterpreted - much better to use something like MJDs which are less likely to be automatically parsed by iso-8601 compliant libraries which will get it wrong unless you live in the UK in the winter…. 

Regards,
	Paul.