- From: Andreas Tolfsen <ato@mozilla.com>
- Date: Wed, 11 Feb 2015 13:22:42 +0100
- To: "public-browser-tools-testing@w3.org" <public-browser-tools-testing@w3.org>
On 11 Feb 2015, at 11:20, James Graham <james@hoppipolla.co.uk> wrote: > If `DELETE /session/{session_id}/window_handle` removes the last window > the remote end is supposed to close the session. However there is no > response to indicate to the local end that this happened. Adding such a > response would allow the local to take whatever steps are required to > indicate that there is no longer a session, rather than retaining stale > session information. The assumption is that any subsequent commands to the session after it's ended should yield errors. Because DELETE /session/{id} returns an OK response when the session is successfully ended and an error response when not, the local end can tell when to clean up local resources or take other steps. It would be useful to give the local end an idea of whether the session was ended as a result of calling another command. A tangent is that we also operate with fatal and non-fatal errors as responses to commands, but we're not explicit about which is which. What are you suggesting we do to indicate that an OK response from closing the last window also implicitly ended the session? Since our protocol is unidirectional we need to embed that somehow into the response. Perhaps an extra boolean field? (Related I found no evidence of the spec actually mentioning that subsequent calls to an ended session should give you errors, so I filed a bug about that: https://www.w3.org/Bugs/Public/show_bug.cgi?id=27998)
Received on Wednesday, 11 February 2015 12:23:13 UTC