Terribly sorry to have broken the view. On balance I thought it might be quicker to go forwards rather than back - but if it will take too long to switch over to IDs I can try to back out my change and investigate what it changed in terms of the request sent to CouchDB. The Couch URL which is being used "looks reasonable" in that it is hard to see how it might be different to the URL which was being sent before the change but presumably there is some subtle difference in the exact encoding which is being applied to the JSON data.

That said, if we *do* go forwards, I think it would be important to use a stable domain-specific key rather than the doc UUID. It would be impossible to operate stable URLs for artefacts and other entities otherwise. Hugues has told us how to derive these for McCord and it is an essential requirement of any other integration that we could do this for another institution. So this is certainly something we will do for 0.5 but the question is whether it makes sense this week.

Sorry for the regression,
Antranig.

On 11/02/2010 09:22, Colin Clark wrote:
Hey,

Couch UUIDs are probably the only stable identifier we can use without making 
assumptions about the structure of the museum's data. It makes sense to me to 
go ahead and do this, though I would have perhaps preferred postponing that 
change until after the 0.3b release for Hugue's pilot.

Antranig, I think your commit did indeed break the Exhibition About page. The 
question I have is how we should fix this; Yura, you mentioned that you think 
the switch from keying by title to UUID is trivial. Do we know if this change 
will also fix Exhibition About?

Regressions, regressions. :(

Colin

On 2010-02-11, at 3:43 AM, Antranig Basman wrote:

Could I suggest in the meantime that instead of issuing titles in URLs we 
instead use document UUIDs, which will head off risks of encoding issues of 
this kind? Titles are not a stable element of the data domain.
In truth, we should be deriving a proper domain-specific stable key, but I 
don't know enough about the McCord set to find this right now.

---
Colin Clark
Technical Lead, Fluid Project
http://fluidproject.org


_______________________________________________________
fluid-work mailing list - [email protected]
To unsubscribe, change settings or access archives,
see http://fluidproject.org/mailman/listinfo/fluid-work

Reply via email to