- From: Gregory Williams <greg@evilfunhouse.com>
- Date: Mon, 24 Sep 2012 15:10:30 -0400
- To: SPARQL Working Group <public-rdf-dawg@w3.org>
I talked a bit with Richard Cyganiak off-list last week about his outstanding comment (RC-2) regarding error messages in the Protocol spec. I tried to explain the reasoning behind our recent desire to use non-normative language to suggest the use the HTTP status message line to indicate the type of error (while otherwise leaving the format of error messages un-specified). He is not happy with this approach, suggesting that limitations of existing software makes the use of the http status message difficult or impossible in many cases. He is still suggesting that the protocol normatively recommend (but not require) that error messages be returned as text/plain, encouraging interoperability of tools that need to produce and consume errors generated by endpoints. We've talked about this before without reaching consensus. Would WG members support Richard's suggested changes on normatively recommending the use of text/plain errors? thanks, .greg
Received on Monday, 24 September 2012 19:10:55 UTC