Change match syntax

[odersky]

match is now treated like an alphabetic operator, it can be appear infix as
well as after ..

Motivation given in https://contributors.scala-lang.org/t/pre-sip-demote-match-keyword-to-a-method/2137/36?u=odersky

[odersky]

Best reviewed by looking at changed files directly, since there was a bit of going back of forth on the way.

[lihaoyi]

Since match now behaves for all intents and purposes like a method call, are we going to demote it from a keyword to an intrinsic method the same way eq, ne, ==, etc. are? That also means you could foo match f where f is a partial/total function, which is something you cannot currently do. Also foo.match _ to eta-expand a function value.

These are all downstream changes that should be called out as a result of this syntax update

Otherwise it would be weird if match was the only keyword that you could use with or without a preceding .

[nafg]

Does this make match redundant with pipe?

…

On Sun, Nov 24, 2019 at 7:34 PM Li Haoyi ***@***.***> wrote: Since match now behaves for all intents and purposes like a method call, are we going to demote it from a keyword to an intrinsic method the same way eq, ne, ==, etc. are? Otherwise it would be weird if match was the only keyword that you could use with or without a preceding . — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#7610>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAYAUCMYM2L65AZLL5TR2LQVMMQ5ANCNFSM4JQ3USLA> .

[odersky]

Making match behave more like a method call is an option. But match will have to stay a reserved word in any case. It is used in a special way in match types and inline matches, neither of which can be expressed as simple method calls. So that weakens the case of making it a method, since it would not be a clean abstraction.

[martijnhoekstra]

Making match behave more like a method call is an option. But match will have to stay a reserved word in any case. It is used in a special way in match types and inline matches, neither of which can be expressed as simple method calls. So that weakens the case of making it a method, since it would not be a clean abstraction.

which IMO is exactly the reason it shouldn't look like a method either.

[liufengyun]

LGTM

[liufengyun]

I fear the selection may break the regularity of Scala syntax. It seems the use case is not important enough to compensate for the loss of regularity.

[hrhino]

So that's why they call it "dotty".