Opening a mod file is not sufficient cause to invalidate in the
workspace, so don't.
Change-Id: I2b7703008528e4469be312165deb17fe6856fd74
Reviewed-on: https://go-review.googlesource.com/c/tools/+/214259
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Minimize the issues at master by not running workspace-level diagnostics
on mod file changes. Once the initial workspace load stabilizes we will
be able to go back to that approach.
Also, a couple of minor changes along the way while debugging.
Change-Id: Ib3510e15171326a1b89f08ef0031a3ef7d9ac4ec
Reviewed-on: https://go-review.googlesource.com/c/tools/+/214257
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
A test variant for a package can only be reloaded by running go/packages
on the non-test variants import path with the -test flag. We need to
cache this import path in order to be able to reload a test package
on-demand.
Also, always ignore test main packages by detecting them in the
metadata.
Fixesgolang/go#36473
Change-Id: I6e5a61c8e73689212303ae09fa3aed4a2ee8762a
Reviewed-on: https://go-review.googlesource.com/c/tools/+/213660
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This change consolidates the FileKind into only the FileHandle.
Previously, it had been set in multiple places, which required users to
pass in a FileKind when fetching a file. This resulted in confusion,
particularly in places when users did not have access to the file kind.
Change-Id: I9e07d7320c46a21d453ffe108d1431a615706a71
Reviewed-on: https://go-review.googlesource.com/c/tools/+/213459
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This is the first step to surfacing potential fixes and suggestions to a user's go.mod file. Specifically, it will show a warning if you have a dependency that is not used, or if a dependency is declared as indirect when it should be direct and vice versa.
This CL adds functionality for version of Go that are >= 1.14.
Updates golang/go#31999
Change-Id: Id60fa0ee201dcd843f62e2659dda8e795bd671db
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211937
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
When the tempModfile flag was enabled, there would be some go
commands that would run without the -modfile flag. This change adds
the config's buildFlags to the invokeGo command in go/packages and
updates how we are checking if the go version is 1.14 within the
modfileFlagExists function.
Fixesgolang/go#36247
Change-Id: I436666c3fcad33e85ba00aec5f227fe4a0e638b8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212477
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
The initial workspace load was happening when a view was created, in serial.
It should really just be kicked off in a separate goroutine once we create a
new view. Implementing this change required some other significant changes,
particularly the additional work being done by the WorkspacePackageIDs
method.
Some other changes had to be made while debugging. In particular, the
modification to the circular dependencies test was a consequence of
golang/go#36265.
Change-Id: I97586c9574f6c4106172d7983e4c6fad412e6aa1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212102
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Test variants and test mains can result in multiple packages being
loaded for a single ID. Handle this case in the PackageHandle function
instead of returning an error.
Change-Id: Ic0024c5bded162a3e78a9cdcb9566449f3683e35
Reviewed-on: https://go-review.googlesource.com/c/tools/+/213457
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
When dealing with cgo files, we need to pass a Mapper for the authored
file. I got most of the sites in CL 208501, but missed this one.
I was nervous about starting to propagate errors, but all the tests
passed, so shrug.
Change-Id: I7505e670e9c01d719c72c3f99d01c8153c3d2ff5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212862
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
No point in constructing the defaults in three places.
Change-Id: I2b0776910a933a7250245bd82dc27e63c34df18a
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212632
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
We have multiple use cases for scanning: goimports, import completion,
and unimported completions. All three need slightly different features,
and the latter have very different performance considerations. Scanning
everything all at once and returning it was not good enough for them.
Instead, design the API as a series of callbacks for each
directory/package: first we discover its existence, then we load its
package name, then we load its exports. At each step the caller can
choose whether to proceed with the package. Import completion can stop
before loading exports, goimports can apply its directory name
heuristics, and in the future we'll be able to stop the scan short once
we've found all the results we want for completions.
I don't intend any significant changes here but there may be some little
ones around the edges.
Change-Id: I39c3aa08cc0e4793c280242c342770f62e101364
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212631
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
The metadata for the workspace packages may not be available when we
need it, so we should allow loading a single package ID. This can be
improved in follow-up CLs by consolidating the individual IDs into one
call to packages.Load. Some adjustments from CL 212102 were split out
into this CL.
Change-Id: I173a79a3cb136530bc99d093f1c2be189eac8ce2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212628
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This commit will delete the temporary go.mod file that is attached
to a view when the session is shutdown.
Updates golang/go#31999
Change-Id: I0f6b0663b0a814ce45d9b12468669f415f8c36aa
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212479
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
CL 212037 introduced a bug with saving overlays. Since VS Code sends nil
contents for a file on save, we were deleting overlay contents on save.
This resulted in very strange behavior.
Also rename overlay.data to overlay.text to match variable names.
Fixesgolang/go#36224
Change-Id: I7f2d12e369aa7f6daa2c9f36c33468ec6bf61930
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212199
Reviewed-by: Michael Matloob <matloob@golang.org>
In one of my previous refactoring changes, I lost the fact that the
context should be detached before invalidating a file's contents. If
this function is canceled, we will be in a bad state.
Also, small change to return ctx.Err() instead of a custom error message
from (*packageHandle).check.
Change-Id: I19e513e09e438feee105fdd89cb7364a0c3c5e7f
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212104
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Now that `go list` errors are sufficient for us to determine a circular
dependency, we don't need to cache errors on dependency packages.
Change-Id: I0633aeb356f93d21afed3371d61d7eae7de255ac
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212197
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rohan Challa <rohan@golang.org>
Reviewed-by: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
This change merges the small helper functions that modified overlays
into a single function and removes the openFiles sync.Map in the view.
Change-Id: Id94c7d86228c9628b7373fab0030ad0c8018dda5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212037
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This change eliminates the extra step of calling GetFile on the view and
getting the FileHandle from the snapshot. It also eliminiates the
redundant source.File type. Follow up changes will clean up the file
kind handling, since it still exists on the fileBase type.
Change-Id: I635ab8632821b36e062be5151eaab425a5698f60
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211778
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
On startup gopls runs go/packages over the entire workspace. The log
message in question outputs each package found along with all the
package's filenames. Obviously in a large project this produces an
incredible amount of output. Fix by putting the log message behind the
"verboseOutput" flag when invoking go/packages in the "dir/..." mode.
I also added the go/packages "query" string to the
once-per-go-packages-call log message so it is more useful.
Change-Id: I651419e34a855325056bca6720eda8671f7d5fa8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210739
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This CL adds a "disableTempModfile" boolean that can be turned on or off.
While we are adding support for go.mod files in gopls, this flag will allow
users to opt out of using the -modfile flag that is enabled in Go 1.14. This
flag might be removed at a future point, the decision still needs to be made.
Updates golang/go#31999
Change-Id: Ic90229333e988fcc6d461ab1ee47bce1114bd965
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212139
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
It wasn't infinite, but gopls would sit at 100% cpu for ~25 seconds
whenever I made a change to a package imported by essentially
everything in my project.
Change-Id: Ifa253a4de06897260e0791888284527258e8de48
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212000
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
The situation in golang/go#35638 was as follows:
didOpen main.go creates a snapshot that knows main.go is in package
"mod.com".
didChange main.go creates a snapshot. When a file changes, we discard
its contents by leaving the file handle out of the "files" map.
didOpen const.go creates a snapshot, and attempts to invalidate the
metadata for packages in the same directory.
The way we detect packages in the same directory is by iterating through
the files in the snapshot. But we threw away the only file in "mod.com"
in step 2 when its contents changed. If a diagnostics run happened to
get in between the two steps, it would re-load main.go and the bug would
go away. If not, step 3 would find no files and fail to invalidate
"mod.com".
The best way to fix this is to insert the new file handle eagerly during
cloning. That way there's no confusion.
Fixesgolang/go#35638.
Change-Id: I340bd28a96ad7b4cc912032065f3c2732c380bb2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211578
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Eliminate the file watcher, since it led to a lot of confusion and
difficulty reasoning about the flow of a file action. This change splits
a file invalidation into the two logical steps - 1) things that affect
the overlay, and 2) things that affect the view. It is based on top of
CL 211757, so the diffs will look better once that CL is merged.
Change-Id: I277475569b61f3c80feaa6b6fe457b4bace82e35
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211777
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This change has no code modifications. Just move the handling for
overlays and debugging into separate files to make them easier to find.
Also, add some missing copyrights.
Change-Id: I7256f704c017457fa3418818d03f89f061af6fc9
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211757
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
In the upcoming Go 1.14 release, there is an introduction of the -modfile
flag which allows a user to run a go command but choose where to direct the
go.mod file updates. The information about this can be found here: golang/go#34506.
This change starts setting up the infrastructure to handle the seperate modfile
rather than keep changing a user's go.mod file. To support versions of Go that are
not 1.14, we run a modified "go list" command that checks the release tags to see
if 1.14 is contained.
Updates golang/go#31999
Change-Id: Icb71b6402ec4fa07e5f6f1a63954c25520e860b0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211538
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Cloning is complicated enough without worrying about concurrency, so
hold the snapshot's lock during the entire process.
Consolidate everything into one function. I don't think that the split
was making it easier to understand, and I was able to see and clean up
some extra complexity once it was all in one place. Let's discuss
options if you think the result is too long.
I don't intend any semantic changes in this CL.
Updates golang/go#35638.
Change-Id: I05c4b28875976293f5fcd56248d9c9e468f85cc6
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211537
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Created an analogous data structure for go.mod files when we parse them
using the golang.org/x/mod package. Gopls can now access the data
within a go.mod file using a parseModHandle and the corresponding
parseModData object. This will help down the road when it is time
to implement the lsp functions for go.mod files.
Updates golang/go#31999
Change-Id: Ibd4d64569bbe3df61b203490b63399d479e7d794
Reviewed-on: https://go-review.googlesource.com/c/tools/+/211303
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
After the addition of golang/go#35964, the import cycle error now
has the import stack attached in the message. This CL parses that
stack and attached the import cycle diagnostic to the import versus
just adding it to the first character of the .go file.
Fixesgolang/go#33085
Change-Id: I6f5f067c338879b898829951236f816aa63d9dfa
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210942
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This change is the next step in unification of text synchronization
methods. The logic really belongs in the internal/lsp/cache package
rather than the internal/lsp package.
Pulled out a function to run diagnostics on a file (diagnostics are still
run async).
Change-Id: I5e237411a02af210ad386b37a6c2aa62ef723567
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210784
Reviewed-by: Heschi Kreinick <heschi@google.com>
When we are processing a go.mod file, we are calling go/packages.load
when we should not be. It will always return 0 packages since it is
not a .go file. This CL adds branching inside each internal/lsp protocol
function and also adds a check in snapshot.PackageHandles for the file type
and returns an error. This will prevent `go list` from running on go.mod files for now.
Updates golang/go#31999
Change-Id: Ic6d0e9b7c81e1f404342b98e10b9c5387adde2ee
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210757
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Say you have foo.go and foo_test.go yielding packages "foo" and
"foo.test". Previously when you changed foo_test.go we would
invalidate the foo.test and foo packages. Invalidating foo is not
necessary since it does not depend on any test files. Furthermore, it
caused problems because nothing would refetch foo's metadata until
foo.go changed, so various things (such as finding implementations in
packages that depend on "foo") would be broken.
Now we only invalidate metadata from packages that contain the
modified file. We only invalidate type info from packages that contain
the modified file, or from such packages' transitive reverse
dependencies.
Change-Id: I23d1af91bcdf22fad4faa1b048afe17ef4e403a1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210460
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
When the go.mod file changes, we should invalidate all the files that are
contained in the package for the mod file. This will allow the files to recheck
their packages in case new packages were added in the go.mod file.
This still does not fix issue where changes to go.mod files do not trigger recalculation of diagnostics.
Updates golang/go#31999
Change-Id: I6d8f1531f5c28ed2dca7fb8ad4ee0317fada787b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210557
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
When a file is changed, we invalidate various cached data so we
re-type check and refetch metadata as needed. Previously when a file
changed we would delete the metadata for all transitive reverse
dependencies. This broke all-packages-in-workspace features since we
could no longer fetch the package handle for packages without
metadata.
Fix by only deleting metadata for the packages that the file being
changed belongs to. It doesn't seem like a package's metadata contains
anything that is sensitive to changes in the package's dependencies.
Change-Id: I6a2d5df49ecd4d627b37689e48ed48fe78ce658d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210458
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
When a new file is opened, the first time we learn about it will be a
didOpen event. We need to invalidate the package's metadata in that
case too.
Fixesgolang/go#35638.
Change-Id: I36c6171e9c959c48ede9e125679e8dd1be7609f4
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210559
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
I had mistakenly forgotten to return a snapshot along with the view.
Fixesgolang/go#36020
Change-Id: I1fc802b8924fccec1d6aaa110640eaed490c3aa1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210215
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
This change will provide a more useful error when you
are self importing a package. It has TODOs in place to propagate the
"import cycle not allowed" error from go list to the user.
Updates golang/go#33085
Change-Id: Ia868a7c688b0f0a7a9689cfda5ea8cea8ae1faff
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209857
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
The invalidateContent function does not acquire a snapshot's mutex to
avoid blocking other work (even though it probably should since it's
only called after a context is canceled). A case was added to iterate
through files when a file is created, and it did not respect the fact
that the snapshot's mutex was not locked, resulting in a concurrent map
read and write. This change makes sure that the access of the snapshot's
files map is guarded by a mutex.
As a follow-up, we should just acquire snapshot.mu in invalidateContent.
Updates golang/go#36006
Change-Id: Idd904ae582055ce786062df50875ac7f0896fd1c
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210199
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
We don't distinguish between genuine errors and context cancellation in
diagnostics, which often results in superfluous logging of these errors.
Avoid spamming the logs with them by checking.
Also, remove the logic for sending undelivered diagnostics. It's a relic
of old bugs and isn't useful.
Change-Id: I7df226539b9b1eb49ab3aae8d7b0a67f59fb3058
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210197
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Apparently I should've had staticcheck on. We were only reading the
metadata in updateMetadata to calculate unused imports, but that's now
at a higher level.
Change-Id: Id3d54fa736062bbbf1c207b8739e87ed5a90293d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/210078
Run-TryBot: Heschi Kreinick <heschi@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Previously, we returned CheckPackageHandles when creating a new view.
Now, return the view's snapshot. Also, add a WorkspacePackageIDs
function in order to run diagnostics on them.
Fixesgolang/go#35548
Change-Id: Ica7e41f5a7aa3ee9413feb4de4ee16e1825db2e1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209418
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
We weren't maintaining our ancestor node list correctly. This caused
us to fail to make AST repairs in certain cases. Now we are careful to
always append to the ancestors list when recursing.
Updates golang/go#34332.
Change-Id: I9b51ec70572170d9f592060d264c98b1f9720fb8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209966
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Previously, we would reload if a user's import list decreased or simply
changed order. This is not necessary. Now, we only re-run if a new import
needs to be loaded.
Updates golang/go#35388
Change-Id: I47874afe773dddb835ac27b18895e7a082950dc7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209057
Reviewed-by: Heschi Kreinick <heschi@google.com>
When searching for implementations we look at all packages in the
workspace. We do a full parse since we need to look for non-exported
types and look in functions for type declarations. However, we always
type check a package's dependencies in export-only mode to save work.
This leads to what I call the "two world" syndrome where you have both
the export-only and full-parse versions of a package in play at once.
This is problematic because mirror objects in each version do not
compare equal.
For example:
-- a/a.go --
package a
type Breed int
const Mutt Breed = 0
type Dog interface{ Breed() Breed }
-- b/b.go --
package b
import "a"
type dog struct{}
func (dog) Breed() a.Breed { return a.Mutt }
---
In this situation, the problem is "b" loads its dependency "a" in
export only mode so it gets one version of the "a.Breed" type. The
user opens package "a" directly so it gets fully type checked and has
a second version of "a.Breed". The user searches for "a.Dog"
implementations, but "b.dog" does not implement the fully-loaded
"a.Dog" because it returns the export-only version of the "a.Breed"
type.
Fix it by always loading in-workspace dependencies in full parse mode.
We need to load them in full parse mode anyway if the user does find
references or find implementations.
In writing a test I fixed an incorrect import in the testdata. This
uncovered an unrelated bug which made a different implementation test
very flaky. I disabled it for now since I couldn't see a fix simple
enough to slip into this commit.
Fixesgolang/go#35857.
Change-Id: I01509f57d54d593e62c895c7ecb93eb5f780bec7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209759
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Lack of context in error messages is making my life difficult. Add
context to a few, refactoring out some duplicate code along the way.
Change-Id: I3a940b12ec7c82b1ae1fc477694a2b8b45f6ff71
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209860
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
As far as I can tell, the code I removed in from load did roughly
nothing -- returning nil metadata didn't suppress type checking as I
think was intended. Throwing away the metadata also created the race in
Pull the check for missing import changes up to PackageHandles, where it
is non-racy and can cause type checking to be skipped. Simplify and
refactor.
Fixesgolang/go#35951.
Change-Id: Id4b32b86569afb36863aaf982616b2b3727b0e83
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209737
Run-TryBot: Heschi Kreinick <heschi@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Sometimes the prefix of the thing you want to complete is a keyword.
For example:
variance := 123
fmt.Println(var<>)
In this case the parser produces an *ast.BadExpr which breaks
completion. We now repair this BadExpr by replacing it with
an *ast.Ident named "var".
We also repair empty decls using a similar approach. This fixes cases
like:
var typeName string
type<> // want to complete to "typeName"
We also fix accidental keywords in selectors, such as:
foo.var<>
The parser produces a phantom "_" in place of the keyword, so we swap
it back for an *ast.Ident named "var".
In general, though, accidental keywords wreak havoc on the AST so we
can only do so much. There are still many cases where a keyword prefix
breaks completion. Perhaps in the future the parser can be
cursor/in-progress-edit aware and turn accidental keywords into
identifiers.
Fixesgolang/go#34332.
PS I tweaked nodeContains() to include n.End() to fix a test failure
against tip related to a change to go/parser. When a syntax error is
present, an *ast.BlockStmt's End() is now set to the block's final
statement's End() (earlier than what it used to be). In order for the
cursor pos to test "inside" the block in this case I had to relax the
End() comparison.
Change-Id: Ib45952cf086cc974f1578298df3dd12829344faa
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209438
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
We were previously returning errors when we failed to load/check a
user's workspace folder, but now we suppress all errors. We shouldn't
disable gopls functionality if something is broken in a user's workspace
folder, rather, we should fall back to the file= queries that will run
when a user edits a file.
Change-Id: Iae05174ca80d2573c0222ac42f29e5556bda0134
Reviewed-on: https://go-review.googlesource.com/c/tools/+/209420
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
