I too am curious how the 'canonical' id will get specified. I'm assuming Louis just 'punted' on this one for now.
However regarding the dates: the dates that seem to matter are not in the format that John mentioned. Updated and PostedTime matter. Can't "dateOfBirth" : "1/1/1975", be an string? We definitely need to resolve the date format thing for these two fields. Seconds since epoch or XSD Dates. Here's the RESTful spec: Dates and timestamps are represented as strings containing AtomPub (RFC3339) format date-time elements; see section 3.3 of [RFC4287]. These are also known as "XSD Dates". In cases where only a day-of-the-year is desired, e.g., a birthday, the year SHOULD be specified as 0000. Which conflicts with the JS A string specifying the time at which this activity took place in milliseconds since the epoch. What is the ultimate storage format? davep On Mon, Jun 30, 2008 at 12:36 PM, John Panzer <[EMAIL PROTECTED]> wrote: > LG, except for the human readable id strings like 'canonical', and that the > dates are in some unknown format (1/1/1999); would be good to document the > format or better yet use xsd dates. > > John Panzer (http://abstractioneer.org) > > On Mon, Jun 30, 2008 at 12:29 PM, Louis Ryan <[EMAIL PROTECTED]> wrote: > >> I've created a JIRA issue with a sample JSON DB patch attached here >> >> https://issues.apache.org/jira/browse/SHINDIG-413 >> >> There are definitely issues with some of the data that I will work on. I >> will also ping on the RESTful discussion list to point folks at this data >> as >> I think it makes sense to use the same data in the spec document and in the >> canonical DB. >> >> I am working on a driver for this structure and I should have a patch ready >> today >> >> On Fri, Jun 27, 2008 at 7:44 PM, Dan Peterson <[EMAIL PROTECTED]> >> wrote: >> >> > Sounds great to me as well, looking forward to seeing it in the sample >> > container. >> > >> > -Dan >> > >> > On Fri, Jun 27, 2008 at 5:45 PM, Kevin Brown <[EMAIL PROTECTED]> wrote: >> > >> > > That would be really useful for ensuring compatibility between PHP and >> > > Java. >> > > It definitely beats the simple xml fetcher thing :) >> > > >> > > On Fri, Jun 27, 2008 at 5:29 PM, Louis Ryan <[EMAIL PROTECTED]> wrote: >> > > >> > > > Hi, >> > > > >> > > > I would like to propose a canonical data implementation for use in >> > > Shindig >> > > > for unit and end-to-end testing based on a JSON structure that we >> check >> > > > into >> > > > SCM. >> > > > >> > > > The idea is pretty simple, define a JSON structure that contains the >> > > > canonical data in the form described in the RESTful spec. Something >> > like >> > > > >> > > > { "people" : < an array of Person JSON objects>, >> > > > "activities" : <an array of Activity JSON object for people>, >> > > > "data" : <an array of app-data for people>, >> > > > "friendLinks" : < a mapping between people id's to simulate friend >> > > links>, >> > > > "dataLinks" : < a mapping between people id's and their app data> >> > > > } >> > > > >> > > > This structure is loaded into a JSON DB driver and used to satisfy >> > > service >> > > > requests. The driver will support CRUD operations on activities and >> > data >> > > in >> > > > line with the spec. >> > > > >> > > > Advantages: >> > > > - Fully exercises the JSON side of the RESTful spec and its binding >> to >> > > > POJOs >> > > > etc >> > > > - Supports simple mutability >> > > > - Its canonical! >> > > > - Shared between all the implementations >> > > > >> > > > Im working on creating some data which Ill send out if folks are >> > > interested >> > > > >> > > > -Louis >> > > > >> > > >> > >> >
