Upgrade to fastparse v2

[olafurpg]

Fastparse v2 came out recently (http://www.lihaoyi.com/post/Fastparse2EvenFasterScalaParserCombinators.html) and it has binary breaking changes with v1.0.0. It doesn't look like fastparse v2 will use a custom package name like fastparse.v2 (https://gitter.im/lihaoyi/fastparse?at=5bd18969ae7be94016b25649) so all libraries using fastparse must either upgrade to v2 or shade the fastparse dependency.

Would it be possible for ScalaPB to shade its fastparse dependency? This would allow downstream users of ScalaPB to choose whether they want fastparse v1 or v2 on their classpath.

[thesamet]

If I shade the default artifact, then I'll be forcing users to have our private shaded copy of fastparse in the classpath (+potentially duplicate code for Scala.js users). If I publish this under another artifact name (say scalapb-runtime-shaded), then dependencies such as scalapb-json4s and scalapb-runtime-grpc will transitively bring in the original scalapb-runtime which depends on fastparse 1 which would cause a mess.

Maybe we should isolate the parsing out to an external library, and provide two implementations one for fastparse 1 and one for fastparse 2. Regardless, I think we'll have fastparse 2 as the default in ScalaPB 0.9.0

[olafurpg]

Fair enough, I renamed the ticket to "upgrade to fastparse v2". No rush on my side, we don't use the text format parser in Scalameta and we have opted to shade fastparse v1.

[raboof]

Looks like this depends on com-lihaoyi/fastparse#215 (which depends on the Scala Native 0.4.0 release)

[raboof]

Fastparse 2 is not available for scala 2.10, so this also depends on #542 (no it doesn't, the plugin doesn't depend on fastparse, only the runtime)

[dgrnbrg]

I would like to have my application use ScalaPB & Ammonite. Unfortunately, Ammonite seems to depend on Fastparse 2, which is blocking this inclusion.

So, +1 for this issue.

[SethTisue]

looks like this should be closed now? (is there a release with this change yet?)

[raboof]

Yes - 0.9.0-M1 should contain this change

[thesamet]

Yes, this went in via b467afa

[thesamet]

Thanks @raboof for this PR!