Possibly related: the last time I checked, there was no widely adopted schema system for JSON, which is completely appalling. That is, it’s not possible to document an HTTP call as “returning a JSON object with field “timestamp” which is an integer and “value” which is a string and no other fields thank you very much” except by writing the english text. My impression is that in the great XML -> JSON switchover, schemas were the baby that was thrown out with the bathwater.
John Clements On Sat, Dec 6, 2014 at 7:37 AM, Matthias Felleisen <[email protected]> wrote: > > On Dec 6, 2014, at 1:08 AM, Alexis King wrote: > > It doesn’t have to be a struct, but the problem is that typed racket > doesn’t know that a value of type String for instance will be of type > JSExpr or not, so you would have to put (assert x jsexpr?) around > everything that you want to treat as a jsexpr. > > > Ah, that’s interesting. I’d seen assert when digging through the docs, but > I didn’t quite realize how it worked at the time. Looking back at it again, > I’m surprised I missed it, but no matter. > > That *does* solve the problem, albeit rather inelegantly. I’d be > interested to see if there’s another way, but I would understand why it > would be difficult to implement a cleaner approach. > > > From 100,000 ft: > > JSON and XML and such things have one of two types: a coarse one that > does not really address the things a programmer thinks and an incredibly > fine-grained one that expresses all of these things. The latter requires a > highly intrinsic, complicated, difficult to use thing [due to the > union/recursive/dynamic nature of the objects], which no real languages > comes with. I conjecture that TR could accommodate just such a type system > -- but that might be a dissertation. I should talk to Sam. > > -- Matthias > > > > ____________________ > Racket Users list: > http://lists.racket-lang.org/users > >
____________________ Racket Users list: http://lists.racket-lang.org/users
