Our logic to detect whether suggested fixes delete unnecessary code did
not take into account suggested fixes that have a command.
Fixesgolang/go#48938
Change-Id: I70d2bebbf7059236525117cc1f454ef726162a96
Reviewed-on: https://go-review.googlesource.com/c/tools/+/355769
Trust: Robert Findley <rfindley@google.com>
Trust: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Also, change the way we test fillreturns to match the other analyzers.
Change-Id: If2124775c583524ff61017452bf065965a6cc97e
Reviewed-on: https://go-review.googlesource.com/c/tools/+/353171
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
I mistakenly used DynamicConfigurationSupported for all capabilities,
not just the workspace/configuration one.
Fixesgolang/go#48600
Change-Id: Ie9b205d89da6e4d110a5310b31fc1ba22f2b5383
Reviewed-on: https://go-review.googlesource.com/c/tools/+/352055
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Add a new "usesgenerics" analyzer to report uses of features related to
generic programming.
This analyzer may be used to temporarily guard other analyzers until
they may be updated to support generic features.
Along the way, update the typeparams API to return the Info.Instances
map, rather than provide indirect access, so that it can be iterated.
Fixesgolang/go#48790
Change-Id: Ia3555524beff6e19f0b9101582205e26e757c8da
Reviewed-on: https://go-review.googlesource.com/c/tools/+/355009
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Tim King <taking@google.com>
This isn't strictly necessary for some of the cases, but it's better to
use it in all cases. Also added a test to ensure that we avoid
(*token.File).Offset in all of gopls--test was probably overkill, but it
was quick to write.
Change-Id: I6dd0126e2211796d5de4e7a389386d7aa81014f0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/353890
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Semantic tokens will now return typeParameters if the client will accept
them. If the user is not using go1.18, then generic code will not type
check and gopls will not recogize type parameters. If gopls has not
been compiled with go1.18 only some uses of type parameters will be
recognized. In any case the code uses internal/typeparams
to see through IndexListExprs.
There is a global parameter semDebug which should be false in
checked-in code. I use it to see how semantic token decisions are
made; it is for debugging. But there are a lot of small code changes
to restrict logging to a few places.
There is also an unexercised stub to fix a bug where some functions
are misclassified as variables. This will be activated in a future CL,
as it will also change the golden test files.
Change-Id: I5107e67ae25e825b0cdc4c1a744877ce97ba609b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/352570
Trust: Peter Weinberger <pjw@google.com>
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
There are some (presumably provisional) new RPCs, PrepareTypeHierarchy,
SuperTYpes, and Subtypes. ServerCapabilities has a corresponding field
TypeHierarchyProvider, and the client has TypeHierarchy which is
a TypeHierarchyClientCapabilities
The name of the return type of ApplyEdit has been changed from
ApplyWorkspaceResponse to ApplyWorkspaceResult.
(its content has not changed, just the name)
srver_gen.go and its generator helper/helper.go needed a small change
to avoid using _ as an argument.
Change-Id: I8fc828e69659e0333af686ec1dd138b0c09c1c62
Reviewed-on: https://go-review.googlesource.com/c/tools/+/353169
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Peter Weinberger <pjw@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
using the typeparams convention.
Change-Id: I42094b75bd43937bdae099473d6241de887456b4
Reviewed-on: https://go-review.googlesource.com/c/tools/+/353289
Trust: Peter Weinberger <pjw@google.com>
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
I'm going to make a follow-up CL to create a source.Offset function
that will always check inRange, and we should use that everywhere
instead of token.Offset, which is pretty prone to panicking.
Fixesgolang/go#48763
Change-Id: Ia81309400d15a28c133f4b3d41c6239231c2532d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/353889
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Add a test for a parameterized slice to make sure that the simplify
slice analysis still finds the simplification.
Change-Id: I20d5f064bcae60c752f0dee53472dd5db0b18a89
Reviewed-on: https://go-review.googlesource.com/c/tools/+/352089
Trust: Suzy Mueller <suzmue@golang.org>
Run-TryBot: Suzy Mueller <suzmue@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
This CL adds a new infertypeargs analyzer, which finds call exprs where
type arguments could be inferred, and suggests a quick fix to simplify
them.
Along the way, may two changes to the supporting frameworks:
- Initialized types.Info.Instances in go/packages
- Fail analysis tests run with suggested fixes if formatting the
resulting source fails.
Change-Id: Ib15e5bd7c26aa293c5fc18a4cff6bc047e9e31d2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/351552
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This change handles the case of the user creating a go.work file
during the gopls session. It also defaults to go.work/gopls.mod being
used outside of experimental mode, so that you don't have to both set
a setting and have a file.
Change-Id: If118cd2fc95c1b5600a6c06217a3b61605b11e28
Reviewed-on: https://go-review.googlesource.com/c/tools/+/342170
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Type parameters don't have default values, so for now, we just skip them
in fillreturns. I think we should still be able to handle functions
whose return values are all concrete types though.
Change-Id: I1fe4c0d46cd564ff71647294d6cc935762bbd9d3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/352910
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
I missed the TODO in undeclaredname to add support for functions, so
really this belongs in that analyzer. This removes a fair bit of code.
However, the type error analyzers don't really work with the go/analysis
testing framework because the suggested fixes are split from the
diagnostics, so I moved the tests into the gopls tests.
Change-Id: I861a7ad531d2732fe698ee0ac46f23ad53b16812
Reviewed-on: https://go-review.googlesource.com/c/tools/+/351333
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
I find the session options page in gopls debug a bit hard to read. This
change aims to improve readability by:
- Sorting non-default options first (and alphabetically by name)
- Use bold text for option name
- Hide the "current value" if it's string representation equals the
default value's string representation
Change-Id: I93606ae788b97e46dc1d3aff420bb58f4c4d9674
Reviewed-on: https://go-review.googlesource.com/c/tools/+/352130
Trust: Pontus Leitzler <leitzler@gmail.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Anonymous struct quick fixes should provide more information on
which struct they will fill. This adds the first fields of an
anonymous struct to the fill title.
Updates golang/go#48563
Change-Id: I42cee2e8b1b9405ac2e2e8cfb8deb2c7710c3056
Reviewed-on: https://go-review.googlesource.com/c/tools/+/351591
Trust: Suzy Mueller <suzmue@golang.org>
Run-TryBot: Suzy Mueller <suzmue@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
All operations within a file should be in terms of token.Pos, until the
value must be returned.
Fixesgolang/go#48492
Change-Id: I2adfcad8b4729386700c08ba4618e033bf7c3db5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/351629
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
A single failed analyzer should not prevent others from reporting their
diagnostics. This is especially relevant as we work on updating
analyzers to support generic code.
Change-Id: If1d958347649a99df92701fc4a4574cb7020596b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/351550
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Add an analyzer that checks for empty interfaces in constraint position,
that could instead use the new predeclared "any" type.
Change-Id: I6c11f74c479c2cba64b3b12e61d70d157f94393b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/351549
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This adds an analyzer that provides suggested fixes for undeclared name
errors on function calls, implementing a stub of the fuction (with an
empty body). As of now this doesn't try to guess returned types but
only parameters.
Generated functions are appended at the end of the file where these type
errors occur.
Updates golang/go#47558
Change-Id: Iaef45ada6b7b73de1fbe42e5f7e334512b65e6c7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/348829
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Peter Weinberger <pjw@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
This CL updates the internal/typeparams package with the new API from CL
349629, switching Info.Inferred to Info.Instances.
Change-Id: I8f98dc39c844ef8891863c08b747f63924ab1c53
Reviewed-on: https://go-review.googlesource.com/c/tools/+/351250
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
There were two panics that cannot happen (meaning they are inconsistent
woith my understanding of the documentation). These are being
replaced by logging, as if they happen, they will happen deterministically
again if the user restarts gopls and does the same thing.
Change-Id: Ia405ad2883cedcf3de6cb1376eedb0f9250c61e0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/350110
Trust: Peter Weinberger <pjw@google.com>
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
If the user changes the extension of a go file while editing,
the existing code complains 'No packages found for open file'.
This error messsag is not helpful and this CL avoids it.
Change-Id: Idf426f1ce8b83b23757f81aa1f322fea86df9d82
Reviewed-on: https://go-review.googlesource.com/c/tools/+/350789
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Peter Weinberger <pjw@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Fix a typo in docs for MarshalArgs in internal/lsp/command.
Change-Id: Ib11a31f27b4ebb03b6eecd996ccf600154cd3cd0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/350551
Reviewed-by: Peter Weinberger <pjw@google.com>
Trust: Peter Weinberger <pjw@google.com>
Trust: Pontus Leitzler <leitzler@gmail.com>
This change adds a command that can be used to see the difference in
API between two gopls versions. It prints out the changes in a way that
can be copy-pasted into the release notes.
Also, only run the copyright test with 1.18. I wanted to do this before
to use filepath.WalkDir, but now it also doesn't work with generic
syntax (it doesn't use packages.Load, so doesn't respect build tags).
Fixesgolang/go#46652
Change-Id: I3670e0289a8eeaca02f4dcd8f88f206796ed2462
Reviewed-on: https://go-review.googlesource.com/c/tools/+/327276
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
As part of the work for implementing method-stub code generation,
this CL updates the function signature of source/command.go's SuggestedFixFunc
so that a command can operate on the entire source.Snapshot to analyze
and change multiple packages and their files.
Updates golang/go#37537
Change-Id: I8430b2189ce7d91d37ab991f87ba368245293e56
Reviewed-on: https://go-review.googlesource.com/c/tools/+/279412
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
This CL pulls in the latest changes from go/internal/gcimporter, while
avoiding breaking the build on older go versions. To help maintain
compatibility with older Go versions while minimizing the diff with the
standard library importer, the internal/typeparams package was
significantly expanded.
I decided to use type aliases in the internal/typeparams package on Go
version >= go1.18, and placeholder types on Go version < go1.18. This
reduces the amount of copying needed in the APIs, though it might not be
the best decision if we ever decide to export this package.
Documentation was also updated to be more concise and specific to the Go
version being used.
In order to actually fix the x/tools Trybot for packages using generics
in the standard library, we need to switch from the 'typeparams' build
constraint to the 'go1.18' build constraint. This means if we make any
additional API changes in go/types we'll have to submit them with a
broken x/tools Trybot and then immediately fix the x/tools build.
Change-Id: Ifa0b1c37b89dc549ee295fa3a959f03deda86e56
Reviewed-on: https://go-review.googlesource.com/c/tools/+/349949
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Enable to hover runes found in basic literals in various forms.
When a rune is found, the hover message provides a summary composed of a
printable version (if it exists) of the rune, its codepoint and its name.
Behaviour varies slightly depending on the basic literal: rune literals
always display the summary when hovered, string literals only display it
when an escaped rune sequence is found to avoid providing unnecessary
information, and finally number literals only when expressed as a
hexadecimal number whose size ranges from one to eight bytes.
Fixesgolang/go#38239
Change-Id: I024fdd5c511a45c7c285e200ce1eda0669a45491
Reviewed-on: https://go-review.googlesource.com/c/tools/+/321810
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
The suggesteds completions are based on a superficial parse of all
the template files in the package. The code errs on the side of too
many suggestions.
Change-Id: If956ad548327be25517878aab70802cf62d42a50
Reviewed-on: https://go-review.googlesource.com/c/tools/+/341649
Trust: Peter Weinberger <pjw@google.com>
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
As per the current specification, the correct token name that represents methods is `method`. The current implementation does not produce the correct set of result and results in methods being incorrectly highlighted. Furthermore, the parameter names of interface methods are treated as `method` which is not true, so a fix has been applied to handle `ast.FuncType` being `parameter`.
Spec: https://microsoft.github.io/language-server-protocol/specifications/specification-3-16/#textDocument_semanticTokens
Change-Id: Id5d2fafe9f948a4a175ae19b4c94cca8f48a4b79
GitHub-Last-Rev: 9a3b65ca4dd3bb0542969c043a2042001c7fcb63
GitHub-Pull-Request: golang/tools#340
Reviewed-on: https://go-review.googlesource.com/c/tools/+/349449
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Peter Weinberger <pjw@google.com>
Reviewed-by: Peter Weinberger <pjw@google.com>
This shows up every now and then--maybe we need a wrapper function
around token.Offset to check the range.
Updates golang/go#48249
Change-Id: I9c60bc7cc61fcfb2f4e8c6963586d8b8fbb21835
Reviewed-on: https://go-review.googlesource.com/c/tools/+/348429
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
When resolving a position to a package we must consider all packages,
including intermediate test variants. This manifests, for example, when
jumping to definition in a package that is imported as a test variant
(see golang/go#47825).
For now, fix this by threading through an 'includeTestVariants' flag to
PackagesForFile. This isn't pretty, but should be a trivially safe
change: the only effect will be to increase the number of packages
considered in FindPackageFromPos. Since we are discussing future changes
to the API for querying packages from the snapshot, now did not seem
like a good time to undertake significant refactoring.
A regtest based on the original issue is included.
This CL is joint with rstambler@golang.org.
Fixesgolang/go#47825
Co-authored-by: Rebecca Stambler <rstambler@golang.org>
Change-Id: I4693ec69b50ed4acd569cff87883769c1edf332b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/347563
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
We use a case-insensitive comparison so this doesn't really
matter but it's a bit confusing that gopls/doc/settings.md and
internal/lsp/source/api_json.go were inconsistent here. I'm assuming
the latter also shows up as user-visible documentation somewhere,
probably in the VSCode plugin.
Change-Id: I14fa5b9d062266b6de0397aafb36e0ad84730752
Reviewed-on: https://go-review.googlesource.com/c/tools/+/344353
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Peter Weinberger <pjw@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
There is a TODO to remove the "codelens" alias. This updates an
example in the documentation to use the new name.
Change-Id: Id807e99f5ba1da663eae9660fb65521596b0c488
Reviewed-on: https://go-review.googlesource.com/c/tools/+/344352
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
It is unlikely that users want gopls operating on their node_modules
directories, so we should exclude them by default. If a user wants to
include them, they can override their directory filters setting.
This doesn't exclude *any* directory named "node_modules", so we still
need to implement golang/go#46438 to exclude node_modules completely.
Change-Id: I03c42208e62390dc35e44ac5176422ddf8dc53f7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/347297
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
For
import "a/bar/foo"
the existing code just decides the last component is the package name.
But for
import "a/bar/v2" this is incorrect, as the packge name is 'bar'.
The new code uses the result of parsing to derive the package name
from the import string.
That is, the package name was determined syntactically, it is
now determined semantically.
Fixes https://golang.org/issue/47784
Change-Id: Iccdd25e7e3f591e6514b1e0258e9e1879af9ff2d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/343909
Trust: Peter Weinberger <pjw@google.com>
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
I'm not sure why there was a goroutine in the defer here, but I think
this should fix the flakiness of this test. I wasn't able to reproduce
the flake myself locally however, so I couldn't confirm. I'll leave the
issue open until it stops happening.
For golang/go#47508
Change-Id: Ie9fdc68d70fe1634c3ad001441cf3ce0d2693d17
Reviewed-on: https://go-review.googlesource.com/c/tools/+/342674
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
When searching for references or renaming, we start from all packages
containing the current position. But as reported in golang/go#47564,
this fails if we're renaming an object in another package; we need to
start the search from the package containing the object definition.
This CL finds the missing packages by recursively searching all
locations we encounter. For now, this will cause us to consider the
object location, and may also help us behave correctly with respect to
build constraint variants in the future.
While at it, update the regtests to support renaming. This bug could
be exercised with marker tests, but it's good to have a regtest for
renaming anyway.
Fixesgolang/go#47564
Change-Id: I5517e2aeaaa744fcc6b6b96ffbb0b2625b498ed5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/340472
Trust: Robert Findley <rfindley@google.com>
Trust: Suzy Mueller <suzmue@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
In preparation for moving metadata related functionality to a separate
package, move around some types and export some symbols. This is purely
to reduce diffs in subsequent CLs, and contains no functional changes.
Change-Id: I24d4fbd71df78e4d7a84f6598cdf820b41d542a2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/340729
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This will allow us to add more support for the workspace proposal.
Change-Id: Ie557121afe0c16989ac176dc9246d82661a20c44
Reviewed-on: https://go-review.googlesource.com/c/tools/+/341811
Trust: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
The predeclared 'any' type is only valid when completing constraints. We
should support that properly, but for now exclude it from results so
that our completion tests don't fail on Go 1.18.
For golang/go#47669
Change-Id: I7852f844684a6c03da90bf367d45d732e5d1e9bb
Reviewed-on: https://go-review.googlesource.com/c/tools/+/341850
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Add the scope to the command-line-arguments package ID and path so that
multiple command-line-arguments packages can coexist in the workspace.
Fixesgolang/go#47584
Change-Id: Icbfe90d67627f384c54f352e46270ab2bf4240bd
Reviewed-on: https://go-review.googlesource.com/c/tools/+/341611
Trust: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Tweak a few things to reduce garbage:
- Pre-allocate a couple hot error objects in format.go.
- Change methodsAndFields and packageMembers to take a callback
instead of returning a slice.
- Use two queues for breadth first search. This allows us to alternate
between and reuse the queues for each search level instead of
appending to a single queue indefinitely with no reuse.
- Get rid of candidate.names field. This tracked the string name of
each object in the deep completion path. Unlike with DFS before,
due to BFS this has to be copied for every candidate we inspect. Now
we get the object names from each types.Object in candidate.path,
with the addition of a new bitmask field to remember whether each
object needs "()" appended to it.
Using TestBenchmarkFuncDeepCompletion as a benchmark:
name old time/op new time/op delta
Statistics 14.2ms ± 7% 10.3ms ± 1% -27.41% (p=0.016 n=5+4)
name old alloc/op new alloc/op delta
Statistics 4.31MB ± 1% 3.03MB ± 0% -29.60% (p=0.016 n=5+4)
name old allocs/op new allocs/op delta
Statistics 52.7k ± 1% 44.0k ± 5% -16.52% (p=0.008 n=5+5)
Change-Id: I52f619d9a2e8553115be91f05cf8cc5cfa89123e
Reviewed-on: https://go-review.googlesource.com/c/tools/+/323252
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
The lexicalLookup function is used to determine if
a use of an object would be shadowed by a different
definition after a rename. Switch to using LookupParent
which is more careful about the positions of the
identifiers.
Fixesgolang/go#47583
Change-Id: I3dbdf79e537ce637d1276ddbecb094db21f1c26d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/340551
Trust: Suzy Mueller <suzmue@golang.org>
Run-TryBot: Suzy Mueller <suzmue@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
And add a case that can occur while the user is editing the file.
Fix Parsed.LineCol to get a usable answer for the final newline.
Change-Id: I6367ff52051d72431453525279d99d7eb2180703
Reviewed-on: https://go-review.googlesource.com/c/tools/+/339772
Run-TryBot: Peter Weinberger <pjw@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Trust: Peter Weinberger <pjw@google.com>
Unfortunately only after merging CL 338729 did I use it in a multi-view
workspace. That CL added a goroutine per matcher to scan symbols, but
unfortunately did this for each view, resulting in a race if there are
multiple views.
The fix is straightforward.
Change-Id: I405b37921883f9617f17c1e1506ff67b4c661cbc
Reviewed-on: https://go-review.googlesource.com/c/tools/+/340970
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Robert Findley
Rebecca Stambler
pjw
Suzy Mueller
Pontus Leitzler
Francesco Renzi
Marwan Sulaiman
Jean-Hadrien Chabran
Chitoku
Johannes Altmanninger
Muir Manders