Fix #102: Better main class detection #287

smarter · 2017-04-06T11:36:15Z

Previously, the main class detection was handled by
https://github.com/sbt/zinc/blob/1.0/internal/zinc-apiinfo/src/main/scala/xsbt/api/Discovery.scala
which looks for a main method with the correct signature in the
extracted API. This is imperfect because it relies on ExtractAPI
dealiasing types (because Discovery will look for a main method with a
parameter type of java.lang.String and won't recognize
scala.Predef.String), dealiasing means that the extracted API looses
information and thus can lead to undercompilation.

This commit partially fixes this by adding a new callback to AnalysisCallback:

    void mainClass(File sourceFile, String className)

that is used to explicitly register main entry points. This way, tools
do not need to interpret the extracted API, this is much better since it
makes it easier for zinc to evolve the API representation.

This commit does not actually changes ExtractAPI to not dealias, this
can be done in a later PR.

Note that there is another usecase for xsbt.api.Discovery that this PR
does not replace: discovering tests. This is more complicated because
different test frameworks have different ways to discover tests. For
more information, grep for "Fingerprint" in https://github.com/sbt/sbt
and https://github.com/sbt/junit-interface

Note also that the added scripted test does not actually test that this
PR is correct since sbt itself needs to be updated to use this new API,
the branch at
https://github.com/smarter/sbt/commits/fix/main-class-detection does
this, but cannot be merged in sbt until a new zinc is published.

Fix #102

jvican · 2017-04-06T11:40:36Z

This commit does not actually changes ExtractAPI to not dealias, this
can be done in a later PR.

I'm unsure this is possible at all. API detection needs dealiased types to work correctly.

@smarter I'll review this one, thanks for the PR.

smarter · 2017-04-06T11:58:29Z

I'm unsure this is possible at all. API detection needs dealiased types to work correctly.

API detection of what?

jvican · 2017-04-06T12:24:37Z

API detection.

ExtractAPI.scala.

smarter · 2017-04-06T12:28:19Z

Yes, but why do you say that ExtractAPI needs it to work correctly? After #87 and #239 zinc should handle abstract types in the API correctly, so not expanding type aliases should work too.

smarter · 2017-04-06T12:51:06Z

I fixed the added test to:

actually work (I forgot that zinc scripted tests cannot define commands in build.sbt anymore, you have to add them to IncHandler instead)
actually test zinc itself, no matter what sbt uses for main class discovery (I added a checkMainClasses command to IncHandler, that behaves similarly to existing check* commands in there and uses the internal API).

smarter · 2017-04-26T14:53:58Z

Ping for review.

jvican

LGTM. Excellent job!

It's true that we need to put the hooks (or provide the necessary information) to detect main classes for tests. I'll look into the best way to do so.

jvican · 2017-05-05T13:01:06Z

zinc/src/sbt-test/apiinfo/main-discovery/test

@@ -0,0 +1,4 @@
+> compile


These tests look great!

jvican · 2017-05-05T13:01:16Z

internal/zinc-apiinfo/src/main/scala/sbt/internal/inc/ClassToAPI.scala

@@ -94,6 +96,16 @@ object ClassToAPI {
      val defsEmptyMembers = clsDef :: statDef :: Nil
      cmap.memo(name) = defsEmptyMembers
      cmap.allNonLocalClasses ++= defs
+


jvican · 2017-05-05T13:03:16Z

@smarter Sorry, but we merged a formatting PR that invalidates the contents of this one. Can you please update this PR?

@eed3si9n @dwijnand Can you have a quick look to this? I want to merge it. If you don't have time, tell me so and I'll merge it.

eed3si9n

This is imperfect because it relies on ExtractAPI
dealiasing types (because Discovery will look for a main method with a
parameter type of java.lang.String and won't recognize
scala.Predef.String), dealiasing means that the extracted API looses
information and thus can lead to undercompilation.

Is it possible to fix the root cause rather than adding special case for main class detection?

smarter · 2017-05-05T13:57:23Z

Is it possible to fix the root cause rather than adding special case for main class detection?

The root cause is the use of Discovery, the fix is to not use Discovery and the same should be done for tests, and Discovery should be removed. There is no way to make Discovery understand all the semantics of Scala type system.

Previously, the main class detection was handled by https://github.com/sbt/zinc/blob/1.0/internal/zinc-apiinfo/src/main/scala/xsbt/api/Discovery.scala which looks for a main method with the correct signature in the extracted API. This is imperfect because it relies on ExtractAPI dealiasing types (because Discovery will look for a main method with a parameter type of `java.lang.String` and won't recognize `scala.Predef.String`), dealiasing means that the extracted API looses information and thus can lead to undercompilation. This commit partially fixes this by adding a new callback to AnalysisCallback: void mainClass(File sourceFile, String className) that is used to explicitly register main entry points. This way, tools do not need to interpret the extracted API, this is much better since it makes it easier for zinc to evolve the API representation. This commit does not actually changes ExtractAPI to not dealias, this can be done in a later PR. Note that there is another usecase for xsbt.api.Discovery that this PR does not replace: discovering tests. This is more complicated because different test frameworks have different ways to discover tests. For more information, grep for "Fingerprint" in https://github.com/sbt/sbt and https://github.com/sbt/junit-interface

smarter · 2017-05-26T22:39:44Z

Updated, phew! Hopefully this gets merged before the next major reformatting/refactoring of the codebase.

See sbt/zinc#287 for details.

Use seed.next for Success in Cogen for Try

eed3si9n added the ready label Apr 6, 2017

smarter force-pushed the fix/main-class-detection branch from 4e52bf4 to a355a43 Compare April 6, 2017 12:00

smarter mentioned this pull request Apr 6, 2017

Changes to Contraband json files don't trigger invalidation of source that uses the generated *.java code #282

Closed

smarter force-pushed the fix/main-class-detection branch 3 times, most recently from 81bec72 to 1ed5b11 Compare April 6, 2017 12:48

smarter force-pushed the fix/main-class-detection branch from 1ed5b11 to 80a2a13 Compare April 6, 2017 13:06

jvican approved these changes May 5, 2017

View reviewed changes

eed3si9n requested changes May 5, 2017

View reviewed changes

smarter force-pushed the fix/main-class-detection branch from 80a2a13 to 5a4cb09 Compare May 26, 2017 22:12

smarter force-pushed the fix/main-class-detection branch from 5a4cb09 to f10c53c Compare May 26, 2017 22:38

eed3si9n approved these changes May 28, 2017

View reviewed changes

eed3si9n merged commit d253375 into sbt:1.0 May 28, 2017

eed3si9n removed the ready label May 28, 2017

smarter added a commit to smarter/sbt that referenced this pull request Jul 10, 2017

Use the new Zinc main class detection

be187d6

See sbt/zinc#287 for details.

smarter mentioned this pull request Jul 10, 2017

Use the new Zinc main class detection sbt/sbt#3312

Merged

smarter added a commit to smarter/sbt that referenced this pull request Jul 10, 2017

Use the new Zinc main class detection

cfc8a85

See sbt/zinc#287 for details.

cunei pushed a commit to cunei/zinc that referenced this pull request Oct 25, 2017

Merge pull request sbt#287 from ceedubs/cogen-try

28996d8

Use seed.next for Success in Cogen for Try

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Fix #102: Better main class detection #287

Fix #102: Better main class detection #287

Uh oh!

smarter commented Apr 6, 2017 •

edited by eed3si9n

Loading

Uh oh!

jvican commented Apr 6, 2017

Uh oh!

smarter commented Apr 6, 2017

Uh oh!

jvican commented Apr 6, 2017 •

edited

Loading

Uh oh!

smarter commented Apr 6, 2017

Uh oh!

smarter commented Apr 6, 2017 •

edited

Loading

Uh oh!

smarter commented Apr 26, 2017

Uh oh!

jvican left a comment

Uh oh!

jvican May 5, 2017

Uh oh!

jvican May 5, 2017

Uh oh!

jvican commented May 5, 2017

Uh oh!

eed3si9n left a comment

Uh oh!

smarter commented May 5, 2017

Uh oh!

smarter commented May 26, 2017 •

edited

Loading

Uh oh!

Uh oh!

Fix #102: Better main class detection #287

Fix #102: Better main class detection #287

Uh oh!

Conversation

smarter commented Apr 6, 2017 • edited by eed3si9n Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jvican commented Apr 6, 2017

Uh oh!

smarter commented Apr 6, 2017

Uh oh!

jvican commented Apr 6, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

smarter commented Apr 6, 2017

Uh oh!

smarter commented Apr 6, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

smarter commented Apr 26, 2017

Uh oh!

jvican left a comment

Choose a reason for hiding this comment

Uh oh!

jvican May 5, 2017

Choose a reason for hiding this comment

Uh oh!

jvican May 5, 2017

Choose a reason for hiding this comment

Uh oh!

jvican commented May 5, 2017

Uh oh!

eed3si9n left a comment

Choose a reason for hiding this comment

Uh oh!

smarter commented May 5, 2017

Uh oh!

smarter commented May 26, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

smarter commented Apr 6, 2017 •

edited by eed3si9n

Loading

jvican commented Apr 6, 2017 •

edited

Loading

smarter commented Apr 6, 2017 •

edited

Loading

smarter commented May 26, 2017 •

edited

Loading