#barnabywaltersfeel free to expand on use cases for your nested references idea, if you want
#tobiastomit’s not really about my use case. I can for sure build some search, or reuse yours, but still: I think the goal of the JSON format would be to have the data available in a easy to use way.
#tobiastomif you requite some tools, or some own implementation, to get easy stuff done, like get all h-cards, I think it makes it much harder for everyone.
#barnabywalterstobiastom: agreed, and the current form is the most straightforward for the most common use cases
#barnabywaltersrealistically there’s always going to be a whole bunch of work which needs doing to raw canonical JSON before it can actually be used
#tobiastommy use case is not really related to that, I’m just trying to get a summary for an URL right now, which will support microformats, og and others, but this just came to my mind while implementing it.
#barnabywaltersauthorship data, filling in missing data, normalising things, making sure datetimes aren’t in the future, etc
#barnabywaltersso the canonical JSON is a first step towards that, allowing us to make parsers interoperable, and build tools which enhance that canonical data
#tobiastomdo you have a link to the comment-presentation thing?
#barnabywaltersone of the things I’ll be extracting from my feed reader work into it’s own package is a toolkit which will do as much of this normalizing work as possible
#tobiastomyeah, I see your valid point about correcting stuff and so on, but I think having an optimized thing, and not needing an optimizer in the first place would be a better start.
#tobiastomwe are generating the JSON from a source anyways.
#barnabywalterstobiastom: in theory that’s correct, however it requires the parsers to try to understand the semantics of what they’re parsing, not just the structure
#barnabywaltersonce improved toolkits are made, users will not have to worry about the canonical JSON stage
#barnabywaltersbasically only parser developers should have to worry about it
#barnabywaltersbut it’s necessary a) as a way of deterministically comparing and testing parsers, and b) as a standard representation on top of which better tools can be created
#tobiastomagreed on the parser developer point. but I think the main question is: do we want to keep the space for optimizers open on the future, or should the parsers generate that in the first place.
#tobiastomif we want them to generate the optimize, why create a intermediate format in the first place. we just need to test the optimized one.
#tobiastomif I would like to implement a parser now, I would be required to create the JSON structure now, and then the optimized version. That’s more work for me, so why do it?
#barnabywalterstobiastom: because at the moment parsing is deterministic and standardized, but optimising is still open for experimentation
#barnabywaltersoften because optimisation is somewhat context-sensitive, and programming-language specific
#barnabywalterse.g. the level of HTML sanitization required in different cases will vary
#barnabywaltersif it was to be standardised, we would have to figure out each use case, and name each level of sanitization, and document it thoroughly
#tobiastomYeah, I think you are right. My intention is just to generic right now. For me the best case would be if all the different solutions (like og, rdf, whatever) would sit on a table an talk about the optimized format, but I assume that will not happen. :)
#barnabywaltersalso, it’s actually *less* work over all to have the canonical “raw” JSON format, because then parser developers only need to worry about making that, and other people can work on normalizers — and one normalizer will work with any parser
#barnabywalterstobiastom: TBH I don’t think that “optimised format” exists
#barnabywalterstobiastom: it’s understandable. I’ve spent a *lot* of time yak shaving on these sort of enabling projects and am just now getting to the stage where I can come up with an idea for a project, and start working on business logic right away
#barnabywalterswithout having to build yet more parsers and normalisation tools :)
#tobiastomstill not sure if I put up a pull request for my restructuring. :)
#tobiastomas it would break the project, possibly not. :)
#barnabywalterstobiastom: I suspect your suite is going to be objectively easier to use, so once it’s in use by a couple of parsers (certainly php-mf2, point the mf2py devs at it as well) if it works then you can suggest making it the canonical test suite
#barnabywaltershaving the tests as microformats themselves is a cool idea but awkward to use in practise. simple files are much easier
#tantekbarnabywalters: you'll have to bring that up with glennjones - I think he has an explanation about why the tests marked up with microformats are a good thing
#barnabywalterstantek: AFAIK currently tobiastom’s tests are generated from glennjones’s tests (as they are impressively thorough), so the two could coexist
#barnabywaltersbut if one is significantly easier both to maintain and for developers to use, as I suspect is the case, then IMO that should be the test suite we point parser developers towards
#tobiastomyeah, I can update them anytime with changes from glennjones’s tests.
#tobiastomagain, I only created them in the first place to use the great tests from php and automate stuff.
pfefferle and dwayhs joined the channel
#tantekbarnabywalters: sounds like a good discussion between you guys and glennjones