Datalink Feedback VII: REQUEST

[Patrick Dowler]

Yes, changing DALI is cheap "now" :-)

On 14/04/14 12:40 AM, Markus Demleitner wrote:
> First, as no services or clients are implemented against it, changing
> DALI is cheap.  That's not saying it should be done recklessly, it's
> saying DALI can't break any services (or otherwise all our current
> S-services were broken, as each has one aspect or another not
> compliant with DALI).
>
> Then, my preferred path forward would be a (possibly virtual)
> memorandum stating that mandatory REQUEST should be dropped from DALI
> in one way or another, where in Datalink we briefly mention REQUEST
> isn't required.

No one has yet commented in the bigger fact that REQUEST is the one 
feature that allows implementers the freedom to implement multiple 
capabilities with a single resource path. That was a compromise we 
reached awhile back and it has value to some who have certain style of 
development and/or infrastructure.

There has also been seemingly endless pain caused by optional things 
that people ignore for awhile untl they get bitten, so we have also 
moved more to just required rather than optional. We have all complained 
about such-and-such being optional really makes something hard or 
painful or fragile... REQUEST is there to support implementation freedom 
and extensibility, but that will completely fail is it is not mandatory.


I still havent heard anyone say that they need REQUEST to make something 
work, nor have I heard anyone say the above noted implementation 
flexibility use case should be dropped.

-- 

Patrick Dowler
Canadian Astronomy Data Centre
National Research Council Canada
5071 West Saanich Road
Victoria, BC V9E 2E7

250-363-0044 (office) 250-363-0045 (fax)