better diagnostics when potentially duplicate packages are found #6317

tomerd · 2023-03-23T04:55:36Z

motivation: with the introduction of the regsitry, there is a higher likelyhood of running into duplicate packages

changes: perform a heuristics to try and determine if the packages are duplicate and provide a better diagnostics instead of the standard "duplicate target" one

tomerd · 2023-03-23T04:58:06Z

@neonichu @abertelrud for initial review, still need to add tests, but I think this is putting us in a slightly better footing

current diagnostics having swift-nio from source control and swiftpm-test.swift-nio-8 from registry:

error: multiple targets named 'CNIOAtomics' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'CNIODarwin' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'CNIOLinux' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'CNIOSHA1' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'NIO' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'NIOConcurrencyHelpers' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'NIOFoundationCompat' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'NIOHTTP1' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'NIOTLS' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names
error: multiple targets named 'NIOWebSocket' in: 'swift-nio', 'swiftpm-test.swift-nio-8'; consider using the `moduleAliases` parameter in manifest to provide unique names

with this change:

error: multiple identical targets 'CNIOAtomics', 'CNIODarwin', 'CNIOLinux', 'CNIOSHA1', 'NIO', 'NIOConcurrencyHelpers', 'NIOFoundationCompat', 'NIOHTTP1', 'NIOTLS', 'NIOWebSocket' appear in registry package 'swiftpm-test.swift-nio-8' and source control package 'swift-nio', this may indicate that the two packages are the same and can be de-duplicated by activating the automatic source-control to registry replacement, or by using mirrors

also with this change, for potentially duplicate packages not across scm/registry:

error: multiple identical targets 'CNIOAtomics', 'CNIODarwin', 'CNIOHTTPParser', 'CNIOLinux', 'CNIOSHA1', 'NIO', 'NIOChatClient', 'NIOChatServer', 'NIOConcurrencyHelpers', 'NIOEchoClient', 'NIOEchoServer', 'NIOFoundationCompat', 'NIOHTTP1', 'NIOHTTP1Server', 'NIOMulticastChat', 'NIOPerformanceTester', 'NIOTLS', 'NIOUDPEchoServer', 'NIOWebSocket', 'NIOWebSocketServer', '_NIO1APIShims' appear in package 'swiftpm-test.swift-nio-7' and 'swiftpm-test.swift-nio-8', this may indicate that the two packages are the same and can be de-duplicated by using mirrors

neonichu · 2023-03-23T18:15:21Z

Would it make sense to not even mention the entire list of target names, just to make the message a little more readable? e.g. "'CNIOAtomics' and X more targets"

tomerd · 2023-03-25T01:54:32Z

@swift-ci smoke test

tomerd · 2023-03-25T01:54:41Z

@neonichu this is ready for review

tomerd · 2023-03-25T05:56:26Z

@swift-ci smoke test windows

Sources/PackageLoading/PackageBuilder.swift

tomerd · 2023-03-27T16:54:13Z

@swift-ci smoke test

motivation: with the intriduction of the regsitry, there is a higher likelyhood of running into duplicate packages changes: perform a heuristics to try and determin if the packages are duplicate and provide a better diagnostics instead of the standard "duplicate product" one

tomerd · 2023-03-27T16:59:18Z

@swift-ci smoke test

tomerd · 2023-03-27T18:37:51Z

@swift-ci smoke test windows

tomerd · 2023-03-27T19:56:18Z

windows failures seems unrelated. cc @compnerd

…ftlang#6317) motivation: with the introduction of the regsitry, there is a higher likelihood of running into duplicate packages changes: perform a heuristics to try and determin if the packages are duplicate and provide a better diagnostics instead of the standard "duplicate product" one

compnerd · 2023-03-27T20:11:10Z

@tomerd yeah, @etcwilde is aware of this flake, it occurs on macOS as well it seems.

etcwilde · 2023-03-27T20:32:25Z

@_spi(MainActorUtilities) import _Concurrency
 ^

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:44:14: error: generic parameter 'Success' could not be inferred
    let t2 = Task.startOnMainActor {
             ^

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:44:14: error: generic parameter 'Failure' could not be inferred
    let t2 = Task.startOnMainActor {
             ^

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:44:14: note: explicitly specify the generic arguments to fix this issue
    let t2 = Task.startOnMainActor {
             ^
                 <<#Success: Sendable#>, <#Failure: Error#>>

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:44:19: error: type 'Task<Success, Failure>' has no member 'startOnMainActor'
    let t2 = Task.startOnMainActor {
             ~~~~ ^~~~~~~~~~~~~~~~

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:79:14: error: generic parameter 'Success' could not be inferred
    let t2 = Task.startOnMainActor {
             ^

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:79:14: error: generic parameter 'Failure' could not be inferred
    let t2 = Task.startOnMainActor {
             ^

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:79:14: note: explicitly specify the generic arguments to fix this issue
    let t2 = Task.startOnMainActor {
             ^
                 <<#Success: Sendable#>, <#Failure: Error#>>

C:\Users\swift-ci\jenkins\workspace\swiftpm-PR-windows\swift\test\Concurrency\Runtime\startOnMainActor.swift:79:19: error: type 'Task<Success, Failure>' has no member 'startOnMainActor'
    let t2 = Task.startOnMainActor {
             ~~~~ ^~~~~~~~~~~~~~~~

error: command failed with exit status: 1

Yeah, this was the failure that's been hitting me locally. It has multiple levels of weirdness. First is, why are we picking up swift interfaces for the stdlib. We should be using the swiftmodule. Second is, why are we building swiftinterfaces on Windows? We don't have ABI-stability and library evolution on Windows.

This PR will paper over the issue by emitting private swift interfaces for the stdlib and runtimes, but shouldn't be necessary: swiftlang/swift#64587

@tomerd yeah, @etcwilde is aware of this flake, it occurs on macOS as well it seems.

Yep, and it affects Linux. It affects everything, but on every other test run, and not while in a position to actually debug it.

…) (#6336) motivation: with the introduction of the regsitry, there is a higher likelihood of running into duplicate packages changes: perform a heuristics to try and determin if the packages are duplicate and provide a better diagnostics instead of the standard "duplicate product" one

tomerd requested review from abertelrud and neonichu as code owners March 23, 2023 04:55

tomerd force-pushed the fix/better-diagnostics-for-potentially-duplicate-pacakges branch 2 times, most recently from db170bd to 742aa34 Compare March 23, 2023 05:06

tomerd changed the title ~~[wip] better diagnostics when potentially duplicate packages are found~~ better diagnostics when potentially duplicate packages are found Mar 25, 2023

MaxDesiatov reviewed Mar 26, 2023

View reviewed changes

Sources/PackageLoading/PackageBuilder.swift Outdated Show resolved Hide resolved

MaxDesiatov reviewed Mar 26, 2023

View reviewed changes

Sources/PackageLoading/PackageBuilder.swift Outdated Show resolved Hide resolved

MaxDesiatov reviewed Mar 26, 2023

View reviewed changes

Sources/PackageLoading/PackageBuilder.swift Outdated Show resolved Hide resolved

tomerd added 2 commits March 27, 2023 09:54

tests

20c3cb8

tomerd force-pushed the fix/better-diagnostics-for-potentially-duplicate-pacakges branch from 3bd6e26 to 20c3cb8 Compare March 27, 2023 16:56

MaxDesiatov approved these changes Mar 27, 2023

View reviewed changes

neonichu approved these changes Mar 27, 2023

View reviewed changes

tomerd merged commit 12e124b into swiftlang:main Mar 27, 2023

tomerd mentioned this pull request Mar 27, 2023

[5.9] better diagnostics when potentially duplicate packages are found #6336

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

better diagnostics when potentially duplicate packages are found #6317

better diagnostics when potentially duplicate packages are found #6317

Uh oh!

tomerd commented Mar 23, 2023 •

edited

Loading

Uh oh!

tomerd commented Mar 23, 2023

Uh oh!

neonichu commented Mar 23, 2023

Uh oh!

tomerd commented Mar 25, 2023

Uh oh!

tomerd commented Mar 25, 2023

Uh oh!

tomerd commented Mar 25, 2023

Uh oh!

Uh oh!

Uh oh!

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

compnerd commented Mar 27, 2023

Uh oh!

etcwilde commented Mar 27, 2023

Uh oh!

Uh oh!

better diagnostics when potentially duplicate packages are found #6317

better diagnostics when potentially duplicate packages are found #6317

Uh oh!

Conversation

tomerd commented Mar 23, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tomerd commented Mar 23, 2023

Uh oh!

neonichu commented Mar 23, 2023

Uh oh!

tomerd commented Mar 25, 2023

Uh oh!

tomerd commented Mar 25, 2023

Uh oh!

tomerd commented Mar 25, 2023

Uh oh!

Uh oh!

Uh oh!

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

tomerd commented Mar 27, 2023

Uh oh!

compnerd commented Mar 27, 2023

Uh oh!

etcwilde commented Mar 27, 2023

Uh oh!

Uh oh!

tomerd commented Mar 23, 2023 •

edited

Loading