In my testing, the gopls degraded memory mode (currently set via
"memoryMode": "DegradeClosed") did not save as much memory as expected
due to still type checking all packages in the workspace (even if in
ParseExported mode). It is also annoying to get incomplete results from
references and renaming.
I think we can (and should) fix both problems: don't even consider
packages that aren't 'reachable' via open files, but fully type check
the reverse transitive closure of the packages you're working on. This
CL does exactly that, by swapping out the concept of 'workspace
packages' with 'active packages'.
In testing, this decreased my memory footprint while working on std by
3-4x when compared to normal mode, and 2x when compared to the previous
implementation of DegradeClosed.
It still needs more testing before we move this option out of
experimental.
For golang/go#46902
Change-Id: I1e319d0b1607d344d27e797ce32de057d7a583f9
Reviewed-on: https://go-review.googlesource.com/c/tools/+/336410
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>
Fix the imports text edits by computing it on first principles. This
fixes at least a couple bugs:
- Incorrect handling of positions between \r and \n with windows line
endings.
- Incorrect computation of edits when the imports source prefix is
synthetically terminated with just \n, but the actual source has
\r\n.
This is an unsatisfying solution, necessary because of the interaction
of token.File with file termination (token.File does not capture this
information).
Efforts to fix token.File proved complicated, and this is causing
problems for our users, so I think we should do this for now.
For golang/vscode-go#1489
Change-Id: I235caf3960c7201af93800d65546fbab5c6e3f4b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/319129
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
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>
For array, slice, maps and chan candidates, we now support
transforming them to their element type in completions. For example:
var m map[string]int
var _ int = m<>
At <> we complete to "m[]" because we see that the map value type
matches our expected type.
Fixesgolang/go#46045.
Change-Id: Ibee088550193a53744f93217cc365f67f301ae90
Reviewed-on: https://go-review.googlesource.com/c/tools/+/323451
Run-TryBot: Muir Manders <muir@mnd.rs>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Cherry Mui <cherryyz@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Now we search up to three levels for candidate modifiers. For example,
we can now complete "foo" to "foo()()" (double invocation).
Granted this is rarely useful, but it generalizes and simplifies the
searching we did for dereference modifiers.
Updates golang/go#46045.
Change-Id: Ibf0be8158e16a0a26a6344a346f34af8fe182bb0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/323450
Run-TryBot: Muir Manders <muir@mnd.rs>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Trust: Rebecca Stambler <rstambler@golang.org>
"type mod" refers to agglutinative expressions such as dereference
"*", invocation "()", and slicing "[:]". When considering an object as
a completion candidate, we check whether applying a type mod would
make it a better candidate.
Previously we tracked the type mods we wanted to apply to a candidate
by setting bool fields. Now instead we keep a slice of the type mods.
This has two main advantages:
- The mods are now ordered which will allow us to format candidates
properly when the same mods can appear in different order (e.g.
"<-*foo" or *<-foo").
- We can now record any mod multiple times allowing for "<-<-foo" or
"foo()()".
I changed the formatting code to always create a snippet object since
that made things simpler. I had to tweak a few snippet helper methods
to accept a snippet argument rather than creating a new snippet.
This commit's only functional change is that we no longer show any
type mods in candidate labels. For example, the user will now see
"foo" in the completion popup instead of "*foo". Showing the operators
adds noise to the candidate list, and we didn't display them
consistently.
Updates golang/go#46045.
Change-Id: I3ea7baa1ee2fee80a1f8cfe88cbae1093ae269ba
Reviewed-on: https://go-review.googlesource.com/c/tools/+/323449
Run-TryBot: Muir Manders <muir@mnd.rs>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Trust: Rebecca Stambler <rstambler@golang.org>
In cases like:
var foo func(...interface{})
var one int
var two func() (int, int)
foo(<>)
At <> we were preferring "two()" over "one" because we were really
excited that the multi return value function was usable. "one" was not
preferred because the expected value is interface{} (we default to
saying candidates _don't_ match interface{} to give non-type based
aspects of candidate inference a chance to shine).
Fix by applying the corresponding interface{} logic to the assignees
checking: ignore the case of completing into func(...interface{})
since all multi return value functions would match.
Fixesgolang/go#46378.
Change-Id: I355daa75e067e8b14508ca50b8d3b6b727df5fec
Reviewed-on: https://go-review.googlesource.com/c/tools/+/323509
Run-TryBot: Muir Manders <muir@mnd.rs>
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>
For example:
type foo struct { i int }
type bar struct { i int }
var _ foo = <>
Previously at <> we would offer "bar{}" in addition to "foo{}" because
we knew "bar" was convertible to "foo". The "bar{}" completion wasn't
valid as inserted, however, because we didn't include the type
conversion. Anyway, it doesn't make sense to include convertible types
since the directly matching type should also be available. Fix by
tweaking the literal candidate code to skip candidates requiring
conversion.
Fixesgolang/go#46113.
Change-Id: Ica5e4af7f2f8bb2f7e2361efe39694f8bb738c97
Reviewed-on: https://go-review.googlesource.com/c/tools/+/319509
Run-TryBot: Muir Manders <muir@mnd.rs>
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>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Michael Knyszek <mknyszek@google.com>
With the new ParseExported logic, we can lose some unexported fields on
exported structs. This can lead to misleading or malformatted hover
information.
Fix this by ensuring we always extract the Spec from a full parse. Since
this path is only hit via user-initiated requests (and should only be
hit ~once per request), it is preferable to do the parse on-demand
rather than parse via the cache and risk pinning the full AST for the
remaining duration of the session.
For golang/go#46158
Change-Id: Ib3eb61c3f75e16199eb492e3e129ba875bd8553e
Reviewed-on: https://go-review.googlesource.com/c/tools/+/320550
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>
Despite the name, ParseExported only hollowed out declarations -- it
didn't actually drop any from the AST. This leaves a fair amount of
unexported crud behind. Unfortunately, there are a *lot* of ways to
expose an unexported declaration from an exported one, and it can be
done across files. Because of that, discarding unexported declarations
requires a lot of work.
This CL implements a decent attempt at pruning as much as possible from
the AST in ParseExported mode.
First, we analyze the AST of all the files in the package for exported
uses of unexported identifiers, iterating to a fixed point. Then, we
type check those ASTs. If there are missing identifiers (probably due to
a bug in the dependency analysis) we use those errors to re-parse. After
that we give up and fall back to the older, less effective trimming. The
pkg type changes slightly to accomodate the new control flow.
We have to analyze all the files at once because an unexported type
might be exposed in another file. Unfortunately, that means we can't
parse a single file at a time any more -- the result of parsing a file
depends on the result of parsing its siblings. To avoid cache
corruption, we have to do the parsing directly in type checking,
uncached.
This, in turn, required changes to the PosTo* functions. Previously,
they operated just on files, but a file name is no longer sufficient to
get a ParseExported AST. Change them to work on Packages instead. I
squeezed in a bit of refactoring while I was touching them.
Change-Id: I61249144ffa43ad645ed38d79e873e3998b0f38d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/312471
Trust: Heschi Kreinick <heschi@google.com>
Run-TryBot: Heschi Kreinick <heschi@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
In cases like:
append([]string{}, foo<>)
we now prefer objects of type string at <>. Previously we had no
preference. In particular, we now try to infer the slice type from the
first append() arg instead of only from the outer context of the
append() call.
Fixesgolang/go#43240.
Change-Id: I59dfa3b18892c5c87fc5ff53109f51f50ee03d26
Reviewed-on: https://go-review.googlesource.com/c/tools/+/316849
Run-TryBot: Muir Manders <muir@mnd.rs>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Heschi Kreinick <heschi@google.com>
Trust: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
ast.NewPackage mutates the input files, making it difficult to avoid
races with our caching model. I had avoided a race resulting from cache
handle cancellation, just to run into another race in multi-session
servers.
But there's no reason to use ast.NewPackage when we only have a single
file. We can just interrogate the file scope wherever needed.
Fixesgolang/go#45868
Change-Id: I521475b51ee3b1c3e408916affecafbc629b0191
Reviewed-on: https://go-review.googlesource.com/c/tools/+/315629
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: Heschi Kreinick <heschi@google.com>
In LSP, CompletionItems can say if they are for deprecated names. This
CL implements that for items where the doc comments contain a line
starting // Deprecated.
Semantic tokens now similarly mark deprecated tokens, but in vscode
the default theme doesn't change the display, and the customization
options seem limited to:
"editor.semanticTokenColorCustomizations": {
"rules": {
// only foreground, bold, underline, italic
"*.deprecated": {"italic": true}
}
},
Change-Id: I93ccc227bf4e1e30a4f23b40da4d2cbafe1cd925
Reviewed-on: https://go-review.googlesource.com/c/tools/+/313509
Run-TryBot: Peter Weinberger <pjw@google.com>
Trust: 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>
Before this change directory names were used "as is" for package completion.
It could lead to invalid suggestions (for example, 'package 1abc' or package 'ab-cd').
This change adds a check whether a directory name can be used in a package path.
If the directory name is invalid, only 'package main' will be suggested.
Otherwise, the directory name will be normalized and will be used as a package name.
Note: normalized directory names contain only lower case letters and digits.
Fixesgolang/go#44680
Change-Id: I4b710f90d1723c512e29dc3c248a1e681f1cd37f
GitHub-Last-Rev: 8ae69f1c6fdf80831e5773bdb3507a8d51a4a0cf
GitHub-Pull-Request: golang/tools#310
Reviewed-on: https://go-review.googlesource.com/c/tools/+/313092
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>
In cases like:
foo.<>
bar = 123
We weren't detecting that the selector preceding <> was a statement.
The above is parsed as "foo.<>bar = 123", so it looks like <> is
contained in an AssignStmt. This matters because we give different
postfix completions depending on whether it is valid to replace the
surrounding selector with a statement or not.
Updates golang/go#45718.
Change-Id: I8f74505b2c8c7060f1e94433904ff0a987d0cc57
Reviewed-on: https://go-review.googlesource.com/c/tools/+/313269
Run-TryBot: Muir Manders <muir@mnd.rs>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Trust: Rebecca Stambler <rstambler@golang.org>
For example:
var b [4]byte
var s []byte = <>
At <> we now prefer "b" and insert as "b[:]".
Fixesgolang/go#40277.
Change-Id: I5fe9d153813dac7218edf31c7c33610130eef9bc
Reviewed-on: https://go-review.googlesource.com/c/tools/+/311169
Run-TryBot: Muir Manders <muir@mnd.rs>
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>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
Don't offer deep completions when completing the name of unimported
packages.
For example:
var _ int = <>
Completing at <> previously would offer an eclectic array of
candidates such as "bits.LeadingZeros()", "time.Now().Day()", or
"zlib.BestCompression", depending on your luck. These candidates stem
from unimported packages candidates such as "bits" which we continue
searching into for deep candidates.
There are two main reasons these deep completions are not useful:
1. They are not dependable. Not all unimported packages are even
searched (it stops as soon as it finds a set number).
2. Fuzzy matching does not work (e.g. typing "bilz" will not filter to
"bits.LeadingZeros" as it does in other cases).
2) could be remedied, but there are so many unimported
package members that I'm not sure it is possible to reduce false
positive deep completions to a satisfactory level.
I also made a couple relevant minor tweaks:
- Fallback sort the unimported packages by path to keep a consistent
order.
- Don't offer unimported packages at all if there is no prefix.
Updates golang/go#43374.
Change-Id: I9fbcde34a3a9e7781568515bddab9da2fc931139
Reviewed-on: https://go-review.googlesource.com/c/tools/+/311069
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
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>
There are 2 different anonymous struct declarations that require
different approaches:
1. var x struct{...}
2. x := struct{...}{}
For the first one we can use the existing solution with a minor update.
However, it returns the wrong docs for the fields of nested structs.
To fix this we need to visit all fields recursively.
The second one is not a generic declaration. So, the simplest solution
is to use the method Snapshot.PosToField.
Fixesgolang/go#43675
Change-Id: I46685e7985cbf2c1c5b1b74ef3cd3a70b920feba
GitHub-Last-Rev: 8a5704c2ecc3f8a007c00c7adcd637e56d99106c
GitHub-Pull-Request: golang/tools#284
Reviewed-on: https://go-review.googlesource.com/c/tools/+/300029
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Bryan C. Mills <bcmills@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>
Move postfix completion functionality behind an experimental option
flag. For now users can enable it by setting
"experimentalPostfixCompletions" or "allExperiments".
I added a RunnerOption so regtest tests can tweak *source.Options. I
didn't refactor the "Experimental" mode to use the new RunnerOption
because I didn't fully understand its purpose.
Change-Id: I75ed748710cae7fa99f4ea6ea117ce245a4e9749
Reviewed-on: https://go-review.googlesource.com/c/tools/+/296109
Run-TryBot: Muir Manders <muir@mnd.rs>
gopls-CI: kokoro <noreply+kokoro@google.com>
Trust: Heschi Kreinick <heschi@google.com>
Trust: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Postfix snippets are artificial methods that allow the user to compose
common operations in an "argument oriented" fashion. For example,
instead of "sort.Slice(someSlice, ...)" a user can expand
"someSlice.sort!". The snippet labels end in "!" to make it clearer
they do something potentially unexpected. The postfix snippets have
low scores so they should not interfere with normal completions.
The snippets are represented (almost) entirely as Go text/template
templates. This way the user can create custom snippets to match their
general preferences or to capture common patterns in their codebase.
There is currently no way for the user to create snippets, but it
could be accomplished with a configuration file, custom LSP command,
or similar.
I started by implementing a variety of snippets to help flesh out the
various facilities needed by the templates. The most interesting
template capabilities are:
- The ability to import packages as necessary (e.g. "sort" must be
imported to call sort.Slice()).
- The ability to generate unique variable names to avoid accidental
shadowing issues.
- The ability to weave LSP snippets into the template. Currently,
only {{.Cursor}} is exposed, which corresponds to the snippet's
final tab stop.
Briefly, these are the postfix snippets in this commit:
- foo.sort => sort.Slice(foo, func(...){}) (slices)
- foo.last => foo[len(foo)-1] (slices)
- foo.reverse (slices)
- foo.range => for i, v := range foo {} (slices/maps)
- foo.append
This snippet inserts a self-assignment append statement when
appropriate, otherwise just an append expression.
- foo.copy creates a copy of a slice
- foo.clear empties out a map
- foo.keys creates slice of keys
- foo().var assigns result value(s) to variables
- foo.print prints foo to stdout
Some of these are probably not very useful in practice, and I'm sure
there are lots of great ones I didn't think of.
Updates golang/go#39507.
Change-Id: I9ecc748aa79c0d47fa6ff72d4ea671e917a2d5d6
Reviewed-on: https://go-review.googlesource.com/c/tools/+/272586
Run-TryBot: Muir Manders <muir@mnd.rs>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This CL updates addErrCheckAndReturn to return "if err != nil { t.Fatal(err) }"
if the conditions for such snippet completion are met.The CL allows for
the following condition:
The enclosing functions takes a type that implements testing.TB which allows
for tests, benchmarks and extensions to the testing object to be completed.
Also, this CL doesn't explicitly check for the Test/Benchmark function
signature so that test helpers can also get the same benefits.
The remaining conditions for the current "if err != nil" checks also apply.
In the future, more testing completions UX can be added.
Fixesgolang/go#43310
Change-Id: I45197ab25610e31fef629394c79cb3792b532e7d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/279488
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Suzy Mueller <suzmue@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
The comparable interface is introduced on the dev.typeparams branch.
Filter it out from gopls completion results so that it doesn't break
tests on the dev.typeparams branch.
Change-Id: Iba22c0980c09e99b454ce9e22813cc3a1f94a90c
Reviewed-on: https://go-review.googlesource.com/c/tools/+/293931
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
For index expressions, optional "make" args, and composite literal
slice/array keys, we were inferring an expected type of int instead of
untyped int. This caused candidate rankings to not be quite right in
general, and in particular, after support for automatic type
conversions was added, the issue manifested as:
var foo []int
var bar int32
foo[ba<>] // completed to "int(bar)" instead of "bar"
Fixesgolang/go#43375.
Change-Id: I6daef7d23b767f296bdbbc8f47f5b2c972ad9b80
Reviewed-on: https://go-review.googlesource.com/c/tools/+/289272
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>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Hyang-Ah Hana Kim <hyangah@gmail.com>
In cases like:
type foo struct { a int; b float64 }
foo{b<>}
We were completing to "foo{int(b: <float64>)}" (the problem being the
nonsensical int() conversion).
The expected type at "<>" is int to allow completions to match "a".
When we pass the *types.Var representing "b" through the candidate
matching machinery, we say "Oh, a float64! I can convert that to my
expected type of int!".
Fix by bailing out of candidate matching early if the candidate is a
composite literal struct field name. Field names aren't really objects
you can do anything to.
Fixesgolang/go#43789.
Change-Id: Ie4dab166973dfcdcb519f864532ead1f792d25a3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/289130
Run-TryBot: Muir Manders <muir@mnd.rs>
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>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Hyang-Ah Hana Kim <hyangah@gmail.com>
In cases like:
foo<> == 100
We weren't preferring floats at <>. Fix the basic type comparison
logic to know that an untyped int is always compatible with a float.
Fixesgolang/go#44066.
Change-Id: I9cf9bac1632178db100c0a5447351be208b4a2af
Reviewed-on: https://go-review.googlesource.com/c/tools/+/289129
Run-TryBot: Muir Manders <muir@mnd.rs>
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>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Hyang-Ah Hana Kim <hyangah@gmail.com>
Autocompleting a import without quotes appends to the completion,
producing this result: `import math "math/"`.
This commit changes to skip completions when typing a import without
quotes, because the users can be typing the alias of the import.
Fixes: golang/go#42748
Change-Id: I7050989f1f90a6720c17f71f338e50fad1f01456
GitHub-Last-Rev: e7b189a04acd8e501d6d7ac944d25de19156d0da
GitHub-Pull-Request: golang/tools#263
Reviewed-on: https://go-review.googlesource.com/c/tools/+/280652
Reviewed-by: 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>
Trust: Hyang-Ah Hana Kim <hyangah@gmail.com>
Trust: Rebecca Stambler <rstambler@golang.org>
When -mod=readonly and GOPROXY=off are set, the newly imported package
is not type-checked with the new import until it is reflected in the
go.mod file. In such cases, we can continue treating the package as
unimported and get symbols through unimported completions.
Fixesgolang/go#43339
Change-Id: I864c2c6738b537093c0670a266e9030af33f2d36
Reviewed-on: https://go-review.googlesource.com/c/tools/+/280095
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>
For example:
func wantsInt64(int64) {}
func wantsDuration(time.Duration) {}
func _() {
for i := range []string{} {
// inserts "i" as "int64(i)"
wantsInt64(i<>)
// inserts "i" as "time.Duration(i)"
wantsDuration(i<>)
}
}
Type converted candidates have a significant score penalty so they
should never outrank a directly assignable candidate, other factors
being equal.
To minimize false positive completions, we don't offer converted
candidates if either:
- The candidate is a deep completion into another package. This avoids
random "float64(somepkg.SomeVar)" popping up when completing a
float64.
- Don't offer to convert ints to strings since that's not normally
what the user wants.
After going back and forth I decided not to include the type
conversion in the candidate label. This means you will just see the
candidate "i" in the completion popup instead of "float64(i)". Type
names add a lot of noise to the candidate list.
I also tweaked the untyped constant penalty to interplay better with
the new type conversion penalty. I wanted untyped constants to be
ranked above type conversion candidates since they are directly
assignable, so I reduced untyped constants' penalty. However, to
continue suppressing untyped constants from random other packages, I
applied a similar heuristic to above where we give a bigger penalty if
the untyped constant is a deep completion from another package.
Fixesgolang/go#42764.
Updates golang/go#39207 (untyped constant penalty tweak may improve
this situation).
Change-Id: I565114b3a1c9628285d07200b9dfc45c61e5d898
Reviewed-on: https://go-review.googlesource.com/c/tools/+/272286
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
Trust: 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>
Previously struct fields completions are ordered alphabetically.
Now it is ordered by its order in definition.
Fixesgolang/go#42626
Change-Id: I8c404205b97fa9f1a5a66a95a84033b6f0f48eae
Reviewed-on: https://go-review.googlesource.com/c/tools/+/270777
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>
Trust: Rebecca Stambler <rstambler@golang.org>
Trust: Robert Findley <rfindley@google.com>
It felt a little redundant to have completion duplicated in the file
names. Happy to revert if you prefer it the original way.
Change-Id: Ibf14a739688d48ecce0b86154b3cbe4807b2f8b1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/266477
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: Heschi Kreinick <heschi@google.com>
I had originally wanted to create a shared code path for hover in all
cases, but hover has a lot more differences from the documentation in
signature help and completion than I expected. You can't use markdown,
and you probably don't want links--it would take a bigger refactor to
extract something that worked for each feature.
Handling the Structured and SingleLine hover setting also doesn't seem
necessary--those settings are really specific to the way the client
presents the hover, which isn't related to signature help or completion.
For completion, all we need is an extra check on the hover kind for the
NoDocumentation option.
Fixesgolang/go#38577
Change-Id: Ib2037906c13f5be26813fcd2c20989e4d1b6c9bd
Reviewed-on: https://go-review.googlesource.com/c/tools/+/266139
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Trust: Rebecca Stambler <rstambler@golang.org>
Noticed these when I enabled the "unusedparams" analyzer. Not really
necessary to fix most of them, but seemed harmless.
Change-Id: I77c6e5f7c79f1abb701cd3332642486c34049736
Reviewed-on: https://go-review.googlesource.com/c/tools/+/263203
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: Heschi Kreinick <heschi@google.com>
Type-checking can be very expensive for large packages and leave no
budget for shallow completions. This change adds a minimum budget of
10ms so we show the user at least some shallow suggestions.
Updates golang/go#41434
Change-Id: If2ef59c3fbdcfa2ebaabb21256cf606a4f9c14d6
Reviewed-on: https://go-review.googlesource.com/c/tools/+/262347
Trust: Danish Dua <danishdua@google.com>
Run-TryBot: Danish Dua <danishdua@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
This changes bumps relevance for modules with major versions >1 so the
results for unimported members are ordered by major version.
Updates golang/go#41800
Change-Id: I64f4a1cf78a101acf4229433d06b5793246962f5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/261084
Run-TryBot: Danish Dua <danishdua@google.com>
Trust: Danish Dua <danishdua@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Workspace packages seem to contain an empty package name when no
package is present, we can just ignore it since it leads to an
irrelevant completion suggestion.
Fixesgolang/go#41746
Change-Id: I4820751ad0f741dd0d29b42485b41bb4c296a169
Reviewed-on: https://go-review.googlesource.com/c/tools/+/259217
Run-TryBot: Danish Dua <danishdua@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Danish Dua <danishdua@google.com>
This change eliminates any special scenarios where we need to call
c.item instead of going through deepSearch by adding support for all the
cases in deepSearch and c.addItem (previously c.item).
Change-Id: Ifb250be54da2f8c7b656475fcafaa38a4e306244
Reviewed-on: https://go-review.googlesource.com/c/tools/+/258858
Run-TryBot: Danish Dua <danishdua@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Trust: Danish Dua <danishdua@google.com>
This change adds calls to c.matcher.score for commment completion so
comments match the prefix (or fuzzy match depending on user settings)
and don't end up giving the user too many irrelevant suggestions.
Change-Id: Ie660f82c491c17d52e68e781a812bf8053e501f8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/258322
Trust: Danish Dua <danishdua@google.com>
Run-TryBot: Danish Dua <danishdua@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This change removes contexts from functions that don't use them. Passing
down context unnecessarily leads to us having to propagate them
everywhere.
Change-Id: I1723721faf4f487b6cc92b9daef3f23747d9cbc1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/258285
Trust: Danish Dua <danishdua@google.com>
Run-TryBot: Danish Dua <danishdua@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Unimported completions are expensive and can use up a large portion of
completion budget just to find initial deep search candidates. This
change moves these expensive operations which search through the module
cache to after normal deep completions so we search through more useful
candidates first.
Fixesgolang/go#41434Fixesgolang/go#41665
Change-Id: I6f3963f8c65c1a97833a35738d2e96420de2f6ee
Reviewed-on: https://go-review.googlesource.com/c/tools/+/257974
Run-TryBot: Danish Dua <danishdua@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Trust: Danish Dua <danishdua@google.com>
This change switches over load and RunProcessEnvFunc to use the
snapshot's modules instead of the view's modURI. These do not seem to
have been the racy parts of CL 257417.
Change-Id: I317a350fc4b0c62d77858455a0e2e61148804ecd
Reviewed-on: https://go-review.googlesource.com/c/tools/+/257969
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: Heschi Kreinick <heschi@google.com>
This change moves package selection to before type checking so we don't
unnecessarily type-check both variants of a package. As a result, exec
time and memory usage for features making calls to GetParsedFile are cut
by half since we only type check either the narrowest or the widest
package.
Change-Id: Ifd076f8c38e33de2bd3509fe17feafccd59d7419
Reviewed-on: https://go-review.googlesource.com/c/tools/+/257240
Run-TryBot: Danish Dua <danishdua@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Danish Dua <danishdua@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This change refactors deep completion to move search path to a candidate
instead of pointing to it in search queue. This removes the idea of
"being in deep completion" and instead a deep candidate is just a
candidate with a path from search root.
Change-Id: Iec6a0f4409c0c8793986d5c1063000205ed43477
Reviewed-on: https://go-review.googlesource.com/c/tools/+/256417
Run-TryBot: Danish Dua <danishdua@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Danish Dua <danishdua@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Currently, we deep search top level candidates as soon as they're found,
which isn't necessarily leading to an overall breadth first search
approach. This change refactors completion to add candidates to search
queue instead and only call deepSearch once at the end.
Change-Id: I712b03c98b57680a1f1c95df2385f7f2ea6e1ed0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/256259
Run-TryBot: Danish Dua <danishdua@google.com>
gopls-CI: kokoro <noreply+kokoro@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Trust: Danish Dua <danishdua@google.com>
This change improves the labels for import suggestions to only show the
last part of the path. Since VSCode fuzzy searches for labels in text
edit, we now return only the last part of path as text edit instead of
replacing the full import path. Just changing label while returning full
path leads to bad user experience.
Closesgolang/go#35877
Change-Id: Ib10e7a3e030dc9b850ff1d9ec8d45240b75b64a3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/255837
Run-TryBot: Danish Dua <danishdua@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>
Trust: Danish Dua <danishdua@google.com>
Rob Findley
Muir Manders
Heschi Kreinick
pjw
Shoshin Nikita
Marwan Sulaiman
Matheus Alcantara
Rebecca Stambler
Chiawen Chen
Danish Dua
phillc