[Incremental] Obviate the need for a reverse mapping from swiftdeps to swift #728
Conversation
|
@swift-ci please test |
davidungar
changed the title
…ping the swift file in the dependency source of the graph nodes
davidungar
changed the title
|
@swift-ci please test |
| } | ||
| } | ||
| self.previous = previous | ||
| self.disappeared = disappeared.sorted {$0.name < $1.name} | ||
| } | ||
|
|
||
| func isANewInput(_ file: VirtualPath) -> Bool { |
There was a problem hiding this comment.
Can this traffic in handles?
There was a problem hiding this comment.
I'll look into it--good thought!
There was a problem hiding this comment.
I think that changing SourceFile to only traffic in handles would be a good idea, but it leaks out into other code and it might be better to keep this PR as narrow as reasonable. I'll address this idea in a future PR.
|
Took a brief pass over this and LGTM. Thanks for fixing up the modeling here. Going to hold off on approving and leave that to @artemcm who can give you a more thorough review. |
Thank you very much for taking a look so quickly. You are welcome. I'll look into your suggestion about using handles. I like it. |
| let currentSet: Set<VirtualPath.Handle> | ||
|
|
||
| /// Handles of the input files in the previous invocation | ||
| private let previous: Set<VirtualPath.Handle> |
There was a problem hiding this comment.
Nit: Would be nice if this was called previousSet to match the above property.
| @@ -269,12 +269,13 @@ extension IncrementalCompilationState.FirstWaveComputer { | |||
|
|
|||
| return inputsToBeCertainlyRecompiled.reduce(into: Set()) { | |||
| speculativelyRecompiledInputs, certainlyRecompiledInput in | |||
| let speculativeDependents = moduleDependencyGraph.collectInputsInvalidatedBy(input: certainlyRecompiledInput) | |||
| let speculativeDependents = moduleDependencyGraph.collectInputsInvalidatedBy(changedInput: certainlyRecompiledInput) | |||
There was a problem hiding this comment.
Thanks for the names cleanup!
|
@artemcm Thank you for the quick review and the kind words. |
|
@swift-ci please test |
|
The macOS CI appears broken right now, could this pull be the culprit? No hurry, just reporting it so it can be fixed eventually, cc @drexin. |
|
I'll take a look!
… On Jun 26, 2021, at 2:15 AM, buttaface ***@***.***> wrote:
The macOS CI appears broken right now <https://ci.swift.org/job/swift-PR-macos/>, could this pull be the culprit? No hurry, just reporting it so it can be fixed eventually, cc @drexin <https://github.com/drexin>.
—
You are receiving this because you modified the open/close state.
Reply to this email directly, view it on GitHub <#728 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADQTUBW2TFUYFJTESZP52DTUWLCVANCNFSM47HDJ74A>.
|
|
swiftlang/swift#38120 Should fix it as soon as I can test and merge it. I had improved a remark the new driver utters when you pass in -driver-show-incremental.
But I had not fixed the lit test--my bad! Sorry. This PR generalizes the pattern-match.
Thanks
- David
… On Jun 26, 2021, at 9:54 AM, David Ungar ***@***.***> wrote:
I'll take a look!
> On Jun 26, 2021, at 2:15 AM, buttaface ***@***.*** ***@***.***>> wrote:
>
>
> The macOS CI appears broken right now <https://ci.swift.org/job/swift-PR-macos/>, could this pull be the culprit? No hurry, just reporting it so it can be fixed eventually, cc @drexin <https://github.com/drexin>.
>
> —
> You are receiving this because you modified the open/close state.
> Reply to this email directly, view it on GitHub <#728 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADQTUBW2TFUYFJTESZP52DTUWLCVANCNFSM47HDJ74A>.
>
|
Obviate the need for a reverse mapping from swiftdeps to swift by keeping the swift file in the dependency source of the graph nodes.
Remove
BidirectionalMapandInputDependencySourceMap.Bump the minor version number of the priors.
Fixes rdar://77998890