- From: Jacob Beard <jbeard4@cs.mcgill.ca>
- Date: Mon, 11 Feb 2013 21:58:16 -0500
- To: Stefan Radomski <radomski@tk.informatik.tu-darmstadt.de>
- Cc: "VBWG Public (www-voice@w3.org)" <www-voice@w3.org>
- Message-ID: <CADywnFj=UOxteAX+OSFx7Djq3fErEfaWLnKJZkpkHBSV5rrq-w@mail.gmail.com>
This is done. Feel free to send more feedback. Best, Jake On Mon, Feb 11, 2013 at 1:42 PM, Stefan Radomski < radomski@tk.informatik.tu-darmstadt.de> wrote: > Hey Jacob, > > there are quite a few tests where you rely on the ecmascript datamodel, > but do not set the scxml attribute accordingly. Its default is platform > specific and we do use the NULL datamodel per default, therefore more tests > than before fail miserably. Would you consider to set the datamodel > attribute? > > Regards > Stefan > > > On Feb 10, 2013, at 6:36 PM, Jacob Beard <jbeard4@cs.mcgill.ca> wrote: > > FYI, I have made the changes you suggested and pushed them to github. > Furthermore, all tests now validate against scxml.xsd. > > If you have any questions or comments, please don't hesitate to ask. > Best, > > Jake > > On Sun, Feb 10, 2013 at 11:12 AM, Jacob Beard <jbeard4@cs.mcgill.ca>wrote: > >> Hi Stefan, >> >> Thanks for the feedback. My hope was that the scxml-test-framework >> could be adapted for other SCXML implementations, so I'm happy to hear that >> this was possible for uscxml. >> >> Your comments are reasonable, and I'll merge them into the test suite. >> Specifically, here are my replies: >> >> On Sat, Feb 9, 2013 at 6:08 PM, Stefan Radomski < >> radomski@tk.informatik.tu-darmstadt.de> wrote: >> >>> Hey Jacob, >>> >>> I re-ran the tests (informal report below) and there are definitely >>> some issues on my part with parallel states (I submitted another respective >>> mail to this list, but it was not yet delivered it seems). There are a >>> couple of points I'd like to point out: >>> >>> - There is no scxml "profile" attribute (anymore?) >>> >> >> True. There was a profile property in Draft 7 of the spec, when I first >> started developing the test suite. I'll take it out. >> >> In fact, I should probably validate all SCXML test documents against >> the latest XSDs included with the spec. There were previously some problems >> with the XSDs, but I believe this has now been fixed. >> >> >>> - The delay[expr] attribute with send is supposed to be in css2 format >>> (numeric value with "s" / "ms" prefix afaict) >>> >> >> True. I'll fix this. >> >> >>> - You have several occurrences of a "getData" ECMAScript function that >>> is specified nowhere >>> >> >> This is a SCION-specific feature. I'll remove or modify these tests. >> >> >>> - I replied to your framework when there were no more pending delayed >>> sends, not after the first stable configuration, this lead to some failures >>> >> >> The test suite assumes that the interpreter will accept an event, >> execute a macrostep, and will reply with the resulting configuration. If >> you like, we can spawn a new thread and have a conversation about whether >> or not this is the best approach. >> >> >>> - You use the type-sensitive "===" operator in your conds, I am not >>> sure what to make of it as I stringified all event data and the issue is >>> pretty much unspecified >>> >> >> I think if the Content-Type header of the HTTP request is >> "application/json", it makes sense for the SCXML interpreter to parse the >> event data as JSON, and expose it as a js object on the _event.data >> property in the ecmascript datamodel. In this case, using strict equality >> would be appropriate. I agree that this is currently underspecified. >> Language for this could perhaps be added to section D.2.1, Basic HTTP >> Event I/O Processor - Receiving Events. >> >> >>> >>> With the exception of delayed sends I do think that your overall >>> approach is very reasonable and we would support the canonical test cases >>> in this format. Our interpreter is obviously still a little rough around >>> the edges (or blatantly buggy with respect to parallel and transition order >>> :). I will rerun the suite as soon as I ironed out the more obvious bugs. >>> In any case, having tests like these is a *tremendous* help to ensure >>> proper implementation. >>> >> >> Super :) >> >> Best, >> >> Jake >> >> > >
Received on Tuesday, 12 February 2013 02:58:59 UTC