Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

x/tools/gopls: changes cause canceled diagnostics for files in unrelated packages #36144

Closed
myitcv opened this issue Dec 14, 2019 · 8 comments
Closed

x/tools/gopls: changes cause canceled diagnostics for files in unrelated packages #36144

myitcv opened this issue Dec 14, 2019 · 8 comments
Labels
FrozenDueToAge gopls Issues related to the Go language server, gopls. Tools This label describes issues relating to any tools in the x/tools repository. WaitingForInfo Issue is not actionable because of missing required information, which needs to be provided.
Milestone
gopls/v0.4.0

Comments

@myitcv
Copy link
Member

myitcv commented Dec 14, 2019

What version of Go are you using (go version)?

$ go version
go version devel +da4d58587e Sat Dec 7 15:57:30 2019 +0000 linux/amd64
$ go list -m golang.org/x/tools
golang.org/x/tools v0.0.0-20191213221258-04c2e8eff935 => github.com/myitcvforks/tools v0.0.0-20191214162001-758e20ac67bf
$ go list -m golang.org/x/tools/gopls
golang.org/x/tools/gopls v0.1.8-0.20191213221258-04c2e8eff935 => github.com/myitcvforks/tools/gopls v0.0.0-20191214162001-758e20ac67bf

Note this version is a temporary fork with fixes as described in #35638 (comment)

Does this issue reproduce with the latest release?

Yes

What operating system and processor architecture are you using (go env)?

go env Output
$ go env
GO111MODULE="on"
GOARCH="amd64"
GOBIN=""
GOCACHE="/home/myitcv/.cache/go-build"
GOENV="/home/myitcv/.config/go/env"
GOEXE=""
GOFLAGS=""
GOHOSTARCH="amd64"
GOHOSTOS="linux"
GOINSECURE=""
GONOPROXY=""
GONOSUMDB=""
GOOS="linux"
GOPATH="/home/myitcv/gostuff"
GOPRIVATE=""
GOPROXY="https://proxy.golang.org,direct"
GOROOT="/home/myitcv/gos"
GOSUMDB="sum.golang.org"
GOTMPDIR=""
GOTOOLDIR="/home/myitcv/gos/pkg/tool/linux_amd64"
GCCGO="gccgo"
AR="ar"
CC="gcc"
CXX="g++"
CGO_ENABLED="1"
GOMOD="/home/myitcv/gostuff/src/github.com/myitcv/govim/go.mod"
CGO_CFLAGS="-g -O2"
CGO_CPPFLAGS=""
CGO_CXXFLAGS="-g -O2"
CGO_FFLAGS="-g -O2"
CGO_LDFLAGS="-g -O2"
PKG_CONFIG="pkg-config"
GOGCCFLAGS="-fPIC -m64 -pthread -fmessage-length=0 -fdebug-prefix-map=/tmp/go-build705453102=/tmp/go-build -gno-record-gcc-switches"

What did you do?

We have a govim test that starts by opening a file and then calling Formatting.

But before calling Formatting we first wait for the initial diagnostics to be sent for the file.

However very occasionally we do not see the initial diagnostics for the file and the test times out (and fails) waiting.

Here is a failing and passing log of this test:

On the occasions that this test fails, changes to go.mod appear to be related. govim currently supports file watching by simulating the file that has changed being open in the editor. For example, with go.mod changes, govim detects the file has changed, signals DidOpen to gopls on the first change, then DidChange thereafter.

As you can see from the failing log above, the calculation of diagnostics for file.go gets almost to the point of publishing, but then its context gets cancelled. That in itself isn't a problem except that the diagnostics are not subsequently sent.

What did you expect to see?

Initial diagnostics to be sent for a file.

What did you see instead?

As above.

cc @stamblerre @heschik @ridersofrohan
@myitcv myitcv added NeedsInvestigation Someone must examine and confirm this is a valid issue and not a duplicate of an existing one. gopls Issues related to the Go language server, gopls. labels Dec 14, 2019
@myitcv myitcv added this to the gopls v1.0 milestone Dec 14, 2019
@gopherbot gopherbot added the Tools This label describes issues relating to any tools in the x/tools repository. label Dec 14, 2019
@myitcv myitcv mentioned this issue Dec 14, 2019
Closed
@heschi
Copy link
Contributor

heschi commented Dec 16, 2019

This seems like a fairly deep-rooted problem. As I understand it, if we're diagnosing package A, and I edit a file in unrelated package B, we're going to cancel the first diagnostics run and then not re-diagnose A. The fact that it was a go.mod file this time may be something of a red herring.

@myitcv
Copy link
Member Author

myitcv commented Dec 28, 2019

FWIW, this is now the main cause of test failures in the govim test suite.

cc @findleyr

myitcv added a commit to myitcvforks/govim that referenced this issue Jan 1, 2020
@myitcv
tests: fix diagnostic assertions in our tests
35d2c3d 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these
myitcv added a commit to myitcvforks/govim that referenced this issue Jan 1, 2020
@myitcv
tests: fix diagnostic assertions in our tests
224ff29 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these
myitcv added a commit to govim/govim that referenced this issue Jan 4, 2020
@myitcv
tests: fix diagnostic assertions in our tests
20e5372 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these
myitcv added a commit to govim/govim that referenced this issue Jan 5, 2020
@myitcv
tests: fix diagnostic assertions in our tests
aad312a 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these
myitcv added a commit to govim/govim that referenced this issue Jan 5, 2020
@myitcv
tests: fix diagnostic assertions in our tests
8edc344 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these
myitcv added a commit to govim/govim that referenced this issue Jan 5, 2020
@myitcv
tests: fix diagnostic assertions in our tests
f50c1fc 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these
myitcv added a commit to govim/govim that referenced this issue Jan 5, 2020
@myitcv
tests: fix diagnostic assertions in our tests
46a4674 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these
myitcv added a commit to govim/govim that referenced this issue Jan 5, 2020
@myitcv
tests: fix diagnostic assertions in our tests
d08d0ad 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 6, 2020
@myitcv
tests: fix diagnostic assertions in our tests
a2e416d 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 6, 2020
@myitcv
tests: fix diagnostic assertions in our tests
cbbd1f4 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 6, 2020
@myitcv
tests: fix diagnostic assertions in our tests
ae1a351 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 6, 2020
@myitcv
tests: fix diagnostic assertions in our tests
8d0b1c3 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 6, 2020
@myitcv
tests: fix diagnostic assertions in our tests
6d8d896 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 6, 2020
@myitcv
tests: fix diagnostic assertions in our tests
6cef7ca 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 7, 2020
@myitcv
tests: fix diagnostic assertions in our tests
5db2029 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 7, 2020
@myitcv
tests: fix diagnostic assertions in our tests
25ce918 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 7, 2020
@myitcv
tests: fix diagnostic assertions in our tests
9d03c33 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 7, 2020
@myitcv
tests: fix diagnostic assertions in our tests
ac309a5 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 8, 2020
@myitcv
tests: fix diagnostic assertions in our tests
b88a881 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 8, 2020
@myitcv
tests: fix diagnostic assertions in our tests
8535810 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 8, 2020
@myitcv
tests: fix diagnostic assertions in our tests
3d0bfb6 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 8, 2020
@myitcv
tests: fix diagnostic assertions in our tests
4acf155 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
myitcv added a commit to govim/govim that referenced this issue Jan 8, 2020
@myitcv
tests: fix diagnostic assertions in our tests (#653)
dfabc26 
This uber PR contains a number of changes to our testscript tests:

* Drop all errlogmatch's looking for assertions where we really should
  not need them. Historically we needed these because the threading
  model of gopls was a bit broken. i.e. despite us notifying of a change
  to a file, unless we waited for diagnostics to be available then we
  couldn't guarantee that, for example, completions would be correct
* Fix up all go.mod files to include go directives so that we don't fall
  foul of golang/go#36144
* Where we do require errlogmatch's, make sure we follow the existing
  semantics per golang/go#36243 and
  golang/go#36340
* Fix up some scripts so that we don't have initial errors in the file
  and hence don't need to soak up initial diagnostics
* Use sleep $DEFAULT_ERRLOGMATCH_WAIT where we expect there to be no
  diagnostic errors
* Only use errlogmatch on diagnostic publications where we need to
  know that a file has been opened, e.g. file watching, or where we
  need to soak up initial diagnostic notifications (see
  golang/go#36243)
* Move all commented-out errlogmatch commands matching on the number of
  errors to the end of each script. We still can't enable these

Add some initial tests that help to verify our expectations around
diagnostics being published by gopls.

Set GOPATH and GOCACHE consistently for the install testscript scripts
to speed them up.

Call t.Parallel in a couple of places to speed up the entire test run.

Add a test (that is skipped) which capture the essence of
golang/go#36144.
@stamblerre stamblerre changed the title x/tools/gopls: initial diagnostics not sent for file x/tools/gopls: changes cause canceled diagnostics for files in unrelated packages Jan 15, 2020
@myitcv
Copy link
Member Author

myitcv commented Jan 16, 2020

@stamblerre I'm minded to close this given the massive number of updates recently, unless you/@heschik would like to keep it open?

myitcv added a commit to govim/govim that referenced this issue Jan 16, 2020
@myitcv
[DO NOT MERGE]: try to reenable test that stresses golang/go#36144
2e3fcdd
@myitcv
Copy link
Member Author

myitcv commented Jan 16, 2020

FWIW I'm re-enabling a test against current master to stress this out a bit: govim/govim#684

@stamblerre
Copy link
Contributor

Yes, I think it's reasonable to close this now that we diagnose entire workspace on every file change. Even if one diagnostics run gets canceled, another one will immediately be started and will complete once the user stops typing.

@myitcv
Copy link
Member Author

myitcv commented Jan 17, 2020

Given the test case originally described in this issue, I think #36463 is actually more relevant:

The initial workspace load must not be canceled

Once we get a stable master (following a conclusion of #36601) we can easily reenable our test that verifies whether things are working/not here.

myitcv added a commit to govim/govim that referenced this issue Jan 17, 2020
@myitcv
[DO NOT MERGE]: try to reenable test that stresses golang/go#36144
488c269
@stamblerre stamblerre added WaitingForInfo Issue is not actionable because of missing required information, which needs to be provided. and removed NeedsInvestigation Someone must examine and confirm this is a valid issue and not a duplicate of an existing one. labels Jan 17, 2020
@stamblerre
Copy link
Contributor

Sounds good. Please follow up here once you get a chance to test this.

myitcv added a commit to govim/govim that referenced this issue Jan 18, 2020
@myitcv
[DO NOT MERGE]: try to reenable test that stresses golang/go#36144
48d3db1
myitcv added a commit to govim/govim that referenced this issue Jan 20, 2020
@myitcv
cmd/govim: try to reenable test that stresses golang/go#36144
00d81b9
myitcv added a commit to govim/govim that referenced this issue Jan 20, 2020
@myitcv
cmd/govim: reenable test that stresses golang/go#36144
459acb4
@myitcv
Copy link
Member Author

myitcv commented Jan 20, 2020

This does indeed appear to be fixed now; we at least have a passing CI! If we see any further issues we can re-open/create a new issue. Thanks again for all the work on fixes @stamblerre!

@myitcv myitcv closed this as completed Jan 20, 2020
myitcv added a commit to govim/govim that referenced this issue Jan 21, 2020
@myitcv
cmd/govim: reenable test that stresses golang/go#36144
844284b
myitcv added a commit to govim/govim that referenced this issue Jan 22, 2020
@myitcv
cmd/govim: reenable test that stresses golang/go#36144 (#684)
96e61d2
@stamblerre stamblerre modified the milestones: gopls/v1.0.0, gopls/v0.4.0 Jul 22, 2020
@golang golang locked and limited conversation to collaborators Jul 22, 2021
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
FrozenDueToAge gopls Issues related to the Go language server, gopls. Tools This label describes issues relating to any tools in the x/tools repository. WaitingForInfo Issue is not actionable because of missing required information, which needs to be provided.
Projects
None yet
Milestone
gopls/v0.4.0
Development

No branches or pull requests
4 participants
@myitcv @stamblerre @gopherbot @heschi