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

x/tools/gopls: multiple failures on windows-arm-zx2c4 #44232

Open
findleyr opened this issue Feb 12, 2021 · 7 comments
Open

x/tools/gopls: multiple failures on windows-arm-zx2c4 #44232

findleyr opened this issue Feb 12, 2021 · 7 comments

Comments

@findleyr
Copy link
Contributor

@findleyr findleyr commented Feb 12, 2021

Looking at our build dashboard for first-class ports, windows-arm-zx2c4 stands out as failing for a long time.

Some of the failures might be due to resource constraints (gopls tests can be very IO and memory heavy), but some are failures related to the filesystem that should be investigated, for example:
https://build.golang.org/log/477a5faf3f4383900706c4b25940a706322b6362

--- FAIL: TestLSP (23.64s)
    --- FAIL: TestLSP/Modules (8.57s)
        lsp_test.go:57: invalid workspace configuration: case mismatch in path "C:\\Users\\GOPHER~1.DES\\AppData\\Local\\Temp\\TestLSP_Modules233674203\\lsp": component "GOPHER~1.DES" should be "Gopher.DESKTOP-6IM5GCA"
        
--- FAIL: TestCaseInsensitiveFilesystem (0.02s)
    view_test.go:50: checkPathCase("C:\\Users\\GOPHER~1.DES\\AppData\\Local\\Temp\\TestCaseInsensitiveFilesystem086638055\\a\\B\\c\\DEFgh\\f.go") = case mismatch in path "C:\\Users\\GOPHER~1.DES\\AppData\\Local\\Temp\\TestCaseInsensitiveFilesystem086638055\\a\\B\\c\\DEFgh\\f.go": component "GOPHER~1.DES" should be "Gopher.DESKTOP-6IM5GCA", wanted error: false

Opening this issue to track getting this builder out of failure, either by deciding to skip these tests or by fixing them.

CC @stamblerre @heschik @pjweinbgo

@gopherbot gopherbot added this to the Unreleased milestone Feb 12, 2021
@bcmills
Copy link
Member

@bcmills bcmills commented Feb 12, 2021

That builder is really slow in certain ways, so it has a habit of exposing tests that either require a lot of resources or make invalid assumptions about timing.

(See #42925, #43120, #43668.)

@stamblerre stamblerre removed this from the Unreleased milestone Feb 12, 2021
@stamblerre stamblerre added this to the gopls/v1.0.0 milestone Feb 12, 2021
@findleyr
Copy link
Contributor Author

@findleyr findleyr commented Feb 12, 2021

Yeah, I suspect we'll need to skip tests that are resource intensive, but those path errors in particular might warrant consideration.

Also, as a point of clarification, this isn't a first class port (see #27689).

@cespare
Copy link
Contributor

@cespare cespare commented Feb 12, 2021

/cc @zx2c4

@stamblerre stamblerre added this to To Do in gopls on-deck Feb 28, 2021
@bcmills
Copy link
Member

@bcmills bcmills commented Jun 15, 2021

The case mismatch failures are still present. It appears that the test is assuming that a particular component is referred to by its long name, but this builder is instead referring to it by its short name.

(Many Windows filesystems allow each path component to be referred to by both an 8.3 name and a longer name. Tests need to be redundant to those variations on Windows, in much the same way that they need to be redundant to PWD symlinks on macOS.)

@bcmills
Copy link
Member

@bcmills bcmills commented Jun 15, 2021

CC @zx2c4

@zx2c4
Copy link
Contributor

@zx2c4 zx2c4 commented Jun 15, 2021

I suspect this is a consequence of TMP/TEMP being set to something long. I'll try setting that to something shorter.

@zx2c4
Copy link
Contributor

@zx2c4 zx2c4 commented Jun 15, 2021

Alright, now it's set to C:\temp, instead of the thing with the 8.3 abbreviations in it. Let's see if that fixes some things.

@stamblerre stamblerre moved this from To Do to P2 in gopls on-deck Aug 12, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
Linked pull requests

Successfully merging a pull request may close this issue.

None yet
6 participants