Hi, On Mon, Mar 16, 2015 at 7:48 PM, Ralph Bean <rbean@xxxxxxxxxx> wrote: > On Sun, Mar 15, 2015 at 03:49:29PM +0530, Ratnadeep Debnath wrote: > I really like part (1). We could pre-process all kinds of data. Keep > statistics and make them all available via API. > >> 2. Render data from saved data. > > I'm not sure how I feel about part (2). It seems to me like it will > introduce a higher overhead to future changes to the service and more > of the same for people trying to use it. By rendering data, I meant render JSON data, and html snippets (pagelets) too. However rendering html snippets would be less flexible. So, I think, we should just go ahead with rendering JSON data. > What if: > > - Someone wants to style their widget differently? > - Combine the data from two widgets into some other widget? > - Use only part of the data typically associated with one widget. > > To do any of those, would they need to submit a patch to our > server-side code, and then have it be reviewed, released and deployed > before they can make use of it? I agree, +1 > How do you feel instead about splitting it up into two pieces: > server-side data denormalization and client side visualization (via a > python lib or maybe better, a js lib?). That way people can query for > the grouped/collated data and then present it however they like. Just > a thought. That's where the REST API comes into picture :) Regards, rtnpro -- Ratnadeep Debnath, https://www.waartaa.com GPG Fingerprint: 033C 8041 A0E9 CDBA 2E02 B785 2119 5486 F245 DFD6 _______________________________________________ infrastructure mailing list infrastructure@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/infrastructure
