Non-side-effecting Read

[peter-empen]

Hi Li,

what is your attitude towards an additional read method that will never throw an exception? I mean a signature like
def maybeRead[T: Reader](s: ujson.Readable): Either[String, T]
where String would be the message of the exception you are throwing.

Although this is easy to achieve on top of the current read, I still wonder whether it was worth to supply.

[lihaoyi]

Hi Peter!

This sounds alike a reasonable either (maybe called readEither?).

One issue with uPickle is I do not think it is sufficiently rigorous with what exceptions are thrown, when and why. Trying to stuff it into an Either or Try would be a good time to start adding tests and documentation for all the error cases, so we know exactly how uPickle behaves in each case

[peter-empen]

Yeah, readEither sounds really good, I'm not opinionated. In Circe decode returns the same.

Is that something you'd do soon or do you prefer me to give it a try?

Would it suffice to check for Exception and extract the massage where ever it was instantiated?

[lihaoyi]

I won’t have time to take a crack at it any time soon, so go ahead

…

On Mon, 1 Jun 2020 at 7:56 PM, peter-empen ***@***.***> wrote: Yeah, readEither sounds really good, I'm not opinionated. In Circe decode returns the same. Is that something you'd do soon or do you prefer me to give it a try? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#312 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAHEB7GXMLETGIT5XC6DU6LRUOJPTANCNFSM4NPW6Q3A> .

[peter-empen]

OK.
To finalize the signature, is it correct to assume that read always fails fast? Otherwise we better use a type that allows to return any number of failures.
And, if we just want to return a single failure, does it make sense to have a ParseFailure class instead of just using String?

[lihaoyi]

yes currently read always fails fast with an exception

[peter-empen]

1. Concerning PR 317, how can I compile against Scala 2.11 in mill locally? I tried upickle.jvm("2.11.12").compile() but it does not work.
2. Also, what is your preferred strategy to replace method calls that were not available in 2.11?

[peter-empen]

@lihaoyi Waiting for your kind review of #317...

[lihaoyi]

Closing this as a wontfix together with #317