Activity
Yesterday
dmitshur reviewed +2 on golang.org/x/build: upgrade go-github and fix pagination to use After instead of Page8h
Thanks.
Updating to go-github v74 SGTM.
To better understand where you ran into this, are you running your own instance of maintnerd configured to mirror that repo, or Go's instance at https://main…Last Week
dmitshur commented on golang.org/x/website/cmd/{adminapp,adminredirect,googlegolangorg}: update App Engine runtime3d
Thanks.dmitshur reviewed +2 on vendor: update x/tools to 3adf0e9, and other repos3d
Thanks. I imagine the optimal long term place to fix this would be in x/tools/cmd/bundle, by adding a feature to it where it includes the copyright comment (if any) from the package it bundles to the…dmitshur reviewed +2 on cmd/dist: make GOROOT unwritable on more builders3d
Thanks. issue14636.go needs fixing to be compatible with unwritable GOROOT.dmitshur reviewed +2 on cmd/internal/testdir: re-enable default all codegen flag on linux-amd643d
Thanks.dmitshur reviewed +2 on all: replace os.Getenv("GO_BUILDER_NAME") with testenv.Builder in tests3d
Thanks.dmitshur reviewed +2 on internal/testenv: update Builder docs with LUCI builder names3d
Thanks.dmitshur commented on vendor: update x/tools to 3adf0e94d
It looks like prior to CL 700996, the x/net/http2 package doc inadvertently had one of the copyright comments merged into the package comment: https://pkg.go.dev/golang.org/x/net@v0.43.0/http2#pkg-ov…dmitshur commented on all: umbrella issue for OOM on windows/arm64 builder4d
Also see the previous related tracking issue #66962.dmitshur reviewed +2 on cmd: delete some more windows/arm remnants5d
@kolyshkin@gmail.com Just noting that this CL has all submit requirements satisfied and since you authored this CL and you have approvers access, you should feel free to go ahead whenever you're read…dmitshur commented on syscall: actually remove unreachable code5d
Thanks.dmitshur commented on syscall: remove unreachable code5d
(1 comment)
From a quick read, I wasn't completely sure whether the unreachable return was being included for some role in the context of this test, and wanted to make a minimal safe change that wou…dmitshur opened a change syscall: actually remove unreachable code5d
CL 702415 claimed to remove unreachable code, but in reality merely hid
it from vet's unreachable pass. Since the unreachable code isn't serving
an active role in the test, do remove it to simplify c…dmitshur reviewed +2 on golang.org/x/tools/gopls/internal/test: update the minimum go command of TestImportAlias5d
Thanks.dmitshur opened a change golang.org/x/tools: empty change5d
DO NOT REVIEW.
DO NOT SUBMIT.
I'm using this CL to try out trybots.dmitshur commented on golang.org/x/build: extend main Go repo's "every vet check is forcibly made high-confidence" behavior to golang.org/x packages that are vendored (or exclude vendored packages from it)5d
Another instance of this happened in https://github.com/golang/go/issues/73998#issuecomment-3275332889. Though in this case it happens that the unreachable check isn't enabled in all.bash, but is bei…dmitshur commented on golang.org/x/tools/go/analysis/unitchecker: TestVetStdlib failures5d
> Separately, I thought a vet failure in stdlib would be reported by all.bash, is that not the case or did something break?
Oh, I see. The unreachable vet check is one of the two explicitly disabled…dmitshur commented on golang.org/x/build/main.star: replace go1.21 with go1.23 in TOOLS_GO_BRANCHES5d
Thanks.dmitshur commented on crypto/internal/fips140: [release-branch.go1.24] revert post-sealing change to FIPS 140-3 module5d
(1 comment)
Yeah, I agree we definitely need a corresponding backport issue for all CLs on Go's release branches after the major release is out.
In order of preference, it's best if a backport CL …dmitshur commented on syscall: remove unreachable code5d
TestASANFuzz failure looks like an unrelated flake. Might be #73979 or possibly a new failure mode. Bypassing and leaving it to watchflakes to report it.dmitshur commented on go/token: annotate if blocks that defeat vet's unreachable pass5d
Thanks.dmitshur commented on syscall: remove unreachable code5d
Thanks.
Thanks, done in CL 702416.dmitshur opened a change go/token: annotate if blocks that defeat vet's unreachable pass5d
Since putting code in an "if true" block is unusual, make it easier
for readers to understand the purpose of doing this.
For #73998.dmitshur commented on golang.org/x/tools/go/analysis/unitchecker: TestVetStdlib failures5d
The original failure from June was fixed by [CL 678317](https://go.dev/cl/678317). (CC @prattmic.)
The test has started to fail again as of Go tip [CL 633515](https://go.dev/cl/633515), which added …dmitshur opened a change syscall: remove unreachable code5d
Reported by go vet:
$ go vet syscall
# syscall_test
# [syscall_test]
syscall/env_unix_test.go:100:4: unreachable code
The TestVetStdlib test in golang.org/x/tools/go/analysis/unitchecker
also r…dmitshur closed an issue crypto/hmac:gofips140-v1.0.0: TestHMACHash/test-10/Clone failures [consistent failure]5d
dmitshur closed an issue crypto/internal/fips140test:gofips140-inprocess: TestVersion failures [consistent failure]5d
dmitshur closed an issue crypto/rsa:gofips140-inprocess: TestModifiedPrivateKey/D+2 failures [consistent failure]5d
dmitshur closed an issue crypto/rsa:gofips140-v1.0.0: TestModifiedPrivateKey/D+2 failures [consistent failure]5d
dmitshur commented on Revert "crypto/internal/fips140: update frozen module version to "v1.0.0""6d
(1 comment)
A CL with this number doesn't seem to exist.dmitshur reviewed +2 on archive/tar: fix typo in benchmark name6d
(1 comment)
That worked, though my +2 now doesn't count since I became the uploader.dmitshur commented on archive/time: fix typo in benchmark name6d
(1 comment)
I also noticed this now, and not too late to fix. :)
```suggestion
archive/tar: fix typo in benchmark name
```
I _think_ I can apply this change for you to save a round-trip. I'll try.dmitshur reviewed +2 on archive/time: fix typo in benchmark name6d
Thanks.dmitshur opened a change golang.org/x/build/main.star: replace go1.21 with go1.23 in TOOLS_GO_BRANCHES6d
CL 605000 kept only the 1.21 branch as a special case for gopls v0.17.0
needs. By now that is no longer needed.
Replace it with release-branch.go1.23 for today's gopls needs.
For golang/go#65917.
F…dmitshur closed an issue : a placeholder issue6d
dmitshur opened an issue : a placeholder issue6d
dmitshur opened an issue golang.org/x/build: "gomin" builders that test golang.org/x modules at their minimum Go toolchain version (using linux/amd64 only to start with)6d
This new-builder issue is a new opportunity enabled by work in #69095: it relies on golang.org/x modules now always having a requirement on a supported Go toolchain version. It is also based on findi…dmitshur commented on golang.org/x/build/main.star: drop GOEXPERIMENT=synctest from Go 1.25+ x/net builders6d
Thanks.dmitshur commented on golang.org/x/build/main.star: drop GOEXPERIMENT=synctest from Go 1.25+ x/net builders6d
> Afterwards, we can probably still stop setting the experiment on Go 1.25+ x/net builders since it's already effectively on.
Done in PS 2, PTAL.dmitshur reviewed +2 on golang.org/x/net/http2: fix race condition when disabling goroutine debugging for one test1w
Thanks. Glad to see this was only this very-small-problem away from working on 1.24 too!
In general it can be handy to exercise the builders that are being fixed, something like:
```suggestion
Chan…dmitshur commented on golang.org/x/build/main.star: stop setting GOEXPERIMENT=synctest in x/net builders1w
Sounds good, it's no problem to do that, but it does mean we'll need the 1.24-only data races fixed then. TryBots couldn't catch it since it involves the intersection of 1.24 and -race mode, which is…dmitshur commented on golang.org/x/website/_content/blog/jsonv2-exp.md: new blog post1w
The blog post will appear on go.dev/blog [automatically](https://go.googlesource.com/website/+/9e3c5c2a0b6c7e21e4969c3d70f786d0fb5501ff/cmd/golangorg/#deploying-to-go_dev-and-golang_org) shortly afte…dmitshur opened a change golang.org/x/build/main.star: stop setting GOEXPERIMENT=synctest in x/net builders1w
CL 642675 started to set GOEXPERIMENT=synctest in x/net builders to
make Go 1.24's experimental testing/synctest available, and in turn
enable getting real-world experience with the package while als…dmitshur commented on archive/tar: optimize nanosecond parsing in parsePAXTime1w
(1 comment)
@mpratt@google.com Just checking my understanding: the reason to suggest adding testing to the benchmark, rather than leaving testing to be done in the [existing test](https://cs.opensou…dmitshur commented on github.com/ebitengine/oto/v3: ability to use oto v3 and v2 in one module broken in oto v3.3.3 (at least on GOOS=darwin)1w
Thank you!
The fact it's now deprecated shows up in the go command, for example when listing available updates:
```
$ go list -m -u github.com/hajimehoshi/oto/v2
github.com/hajimehoshi/oto/v2 v2.4.…dmitshur commented on github.com/ebitengine/oto/v3: ability to use oto v3 and v2 in one module broken in oto v3.3.3 (at least on GOOS=darwin)1w
Yes, similar to [module retraction](https://go.dev/blog/go116-module-changes#module-retraction), you can only make a change like deprecating by publishing a newer version. See https://go.dev/ref/mod#…dmitshur commented on github.com/ebitengine/oto/v3: ability to use oto v3 and v2 in one module broken in oto v3.3.3 (at least on GOOS=darwin)1w
Thanks, that sounds good. I would still encourage you to mark the oto v2 package and module with a "// Deprecated: ..." comment to take advantage that various tooling recognizes and surfaces those co…dmitshur opened an issue github.com/ebitengine/oto/v3: ability to use oto v3 and v2 in one module broken in oto v3.3.31w
I'm not sure about the status of the v2 major version of oto. It's not marked as deprecated, is that because it's still supported alongside v3? Or should it be marked as deprecated because it's unsup…Earlier
dmitshur reviewed +2 on vendor/golang.org/x/tools: update to a09a2fb1w
Thanks.
If you want, you can drop this line before submitting since this change isn't a part of the latest patch set. Not a big deal either way, so up to you.dmitshur commented on vendor/golang.org/x/tools: update to a09a2fb1w
(1 comment)
Looks like the same change was applied in CL 694536. Glad to see they match.dmitshur commented on vendor/golang.org/x/tools: update to a09a2fb1w
(1 comment)
Note that this parent is a bit old (Aug 8), you'll want to rebase past CL 694536 at least, and ideally as far as the latest tip commit available at this time.dmitshur commented on cmd/link/internal/ld: unconditionally use posix_fallocate on FreeBSD1w
Thanks.dmitshur commented on golang.org/x/build/main.star: generalize excluding particularly slow runmod combinations1w
Sounds good, thanks for taking a look.
At this point I'm probably okay with parking this CL for some time. CL 700897 fixed the short-term problem, and this CL shows we have a reasonable path forward…dmitshur commented on golang.org/x/build/main.star: make -longtest-race builders optional in presubmit1w
Yeah. I realize part of my own confusion was when I was thinking of per-runmod presubmit as a boolean, but it's really one of 3 states: "on by default", "on by default if certain filters match", and …dmitshur commented on golang.org/x/build/main.star: make -longtest-race builders optional in presubmit1w
Oh, wait, -longtest _isn't_ on for presubmit for main branch; only for release branches and when filters apply. I see. So that is odd, I'll look.dmitshur opened a change golang.org/x/build/main.star: generalize excluding particularly slow runmod combinations1w
We already track how expensive and how much time extra time runmods
need, so combine host and runmod timeouts for the decision of whether
a builder is too slow to be included in presubmit by default.dmitshur commented on golang.org/x/build/main.star: make -longtest-race builders optional in presubmit1w
I might be misunderstanding this comment or overlooking something, but I think we do use "and" for presubmit already. It's just both -longtest has it on, and -race has it on. Perhaps if you point out…dmitshur commented on golang.org/x/build/main.star: compose presubmit filters via intersection1w
(1 comment)
Sent CL 700897.dmitshur opened a change golang.org/x/build/main.star: make -longtest-race builders optional in presubmit1w
CL 700597 removed presubmit filters for this runmod combination, and we
also want to also take additional action to override their individual
"presubmit on" from being used as is, since the combinati…dmitshur commented on golang.org/x/build/main.star: compose presubmit filters via intersection1w
(1 comment)
In addition to dropping the location filters, we want to also add `includable_only: true` here, right? Otherwise the longtest-race builder is included in presubmit for every CL, even one…dmitshur commented on src/cmd: delete some more windows/arm remnants1w
cmd/go.TestScript/list_symlink_vendor_issue14054 seems to flake at around 10% rate. Tracked in issue #74863. Also a core file left behind, tracked in #69457. Bypassing since they're unrelated.dmitshur reviewed +2 on src/cmd: delete some more windows/arm remnants1w
dmitshur commented on golang.org/x/website/_content/blog/jsonv2-exp.md: new blog post1w
A very nice post. I just left a few minor comments on link mechanics.
In general, for links to other pages on go.dev itself, it's preferable to use relative links to make them work better for multip…dmitshur fixed an issue golang.org/x/build/cmd/relui: error sending release annoucement email1w
We've restored relui's email announcement sending functionality and it seems to be working okay so far:
https://groups.google.com/g/golang-announce/c/PtW9VW21NPs
Closing.dmitshur closed an issue os: `Root.MkdirAll` can return "file exists" when called concurrently on the same path [1.24 backport]1w
Indeed: https://go.dev/doc/go1.25#ospkgos. Closing as this doesn't apply to Go 1.24. Thanks.dmitshur reopened an issue os: volume handles with FILE_FLAG_OVERLAPPED fail when calling ReadAt [1.25 backport]1w
The next step here is to create the backport CL for release-branch.go1.25 (following https://go.dev/wiki/MinorReleases#making-cherry-pick-cls).dmitshur closed an issue runtime: init function run unexpectedly when plugin.Open fails [1.24 backport]1w
See https://github.com/golang/go/issues/75147#issuecomment-3249993701.dmitshur closed an issue runtime: init function run unexpectedly when plugin.Open fails [1.25 backport]1w
We discussed this in a release meeting. This happened 4 releases ago, and it doesn't seem that this bug is preventing programs from working. The fix will be available in Go 1.26, and we agreed not to…dmitshur closed an issue net: `WriteMsgUDPAddrPort` should accept zero value destination addresses [1.24 backport]1w
See https://github.com/golang/go/issues/75003#issuecomment-3249980265.dmitshur closed an issue net: `WriteMsgUDPAddrPort` should accept zero value destination addresses [1.25 backport]1w
We discussed this in a release meeting. The bug has existed since Go 1.18, so it's not very new. Since it's possible to work around it, it seems anyone running into it may have done that. We agreed n…dmitshur commented on os: volume handles with FILE_FLAG_OVERLAPPED fail when calling ReadAt [1.25 backport]1w
@qmuntal Is it sufficient to backport [CL 697295](https://go.dev/cl/697295), or are there more CLs involved?
dmitshur commented on github.com/andygrunwald/go-gerrit: Don't treat + as a special character for queries2w
In general, moving in this direction makes sense to me. This special behavior for '+' and ':' characters was added in PR #22 to resolve issue #18 (and it happened 9 years ago), but the issue describe…dmitshur commented on github.com/XProger/OpenLara: Any documentation about Lara's original movement system?2w
> That "running" animation has own ranges for state switches back to "stop" via another intermediate animation, "jump", "walk" etc. so if you press jump button while running the code start looking an…dmitshur opened an issue cmd/dist: let go in PATH select a newer toolchain to satisfy minimum bootstrap version requirement3w
We regularly bump the minimum Go bootstrap version, once every year (see proposal #54265). One of the common ways that a bootstrap toolchain is provided to make.bash and related scripts is by having …dmitshur commented on golang.org/x/net/quic: TestStreamTransfer failures3w
CC @qmuntal, @golang/windows.dmitshur closed an issue os: File.Seek doesn't set the correct offset with Windows overlapped handles [1.24 backport]3w
dmitshur commented on golang.org/x/crypto/x509roots: regeneration fails with "404 Not Found"4w
Happened again this morning. As of posting this, it reproduces with:
```sh
$ date
Mon Aug 18 12:23:43 EDT 2025
$ go generate ./x509roots
2025/08/18 12:23:45 got non-200 OK status code: 404 Not Found…dmitshur pushed to
main in dmitri.shuralyov.com/service/change4modmitshur commented on dmitri.shuralyov.com/blog: Trying out the Oculus Rift CV1 for the first time1y
July 4, 2024. Today I got a chance to try out the Apple Vision Pro. :)dmitshur pushed to
main in dmitri.shuralyov.com/scratch1ydmitshur created package in dmitri.shuralyov.com/scratch/quine1y
dmitshur merged a change dmitri.shuralyov.com/scratch: quine: add package quine1y
dmitshur reviewed +2 on dmitri.shuralyov.com/scratch: quine: add package quine1y
LGTM. Thanks, and congrats on your first new Go package here!
Done in PS 2.
Patch set 2 seems to resolve this.dmitshur merged a change dmitri.shuralyov.com/scratch: add AUTHORS.md file1y
dmitshur reviewed +2 on dmitri.shuralyov.com/scratch: add AUTHORS.md file1y
Thanks! If it's ready to be submitted from your side, I can do that now.dmitshur pushed to
main in dmitri.shuralyov.com/html/belt1ydmitshur commented on dmitri.shuralyov.com/scratch: main: add package quine1y
They are. Although Go can make them too easy via `go:embed`.
See some initial comments.
I suggest either adding a quite that can be licensed via the existing BSD-3-Clause LICENSE at the top level o…dmitshur reviewed +2 on dmitri.shuralyov.com/scratch: add AUTHORS.md file1y
Thanks! Congrats on your first change. :D
See one inline comment that's unresolved (yellow border). You can leave a draft comment like "Done." and mark the 'Resolved' checkbox, then when you upload …dmitshur pushed to
main in dmitri.shuralyov.com/go/generated2ydmitshur pushed to
main in dmitri.shuralyov.com/gpu/mtl2ydmitshur pushed to
main in dmitri.shuralyov.com/gpu/mtl2ydmitshur commented on dmitri.shuralyov.com/idiomatic-go: Use defer if its overhead is relatively negligible, or it buys meaningful readability3y
I agree with the points made above. This entry tried to say "use this language feature when it's appropriate; don't use it when it's not", which can be said for any language feature. I'll close this …dmitshur pushed to
master in dmitri.shuralyov.com/service/change3ydmitshur closed an issue dmitri.shuralyov.com/go/generated: update implementation for 2020 spec changes3y
dmitshur merged a change dmitri.shuralyov.com/go/generated: update implementation for 2020 spec changes3y
dmitshur commented on dmitri.shuralyov.com/go/generated: update implementation for 2020 spec changes3y
Go fuzzing support is out of beta and no longer uses this build constraint. It should be replaced with `go1.18` by now.
Done in PS 3.
I'm following the Go style for commit messages, which chooses…dmitshur commented on dmitri.shuralyov.com/go/generated: update implementation for 2020 spec changes3y
Shortened 'golang.org' to 'go.dev' here, and on line 24.dmitshur pushed to
master in dmitri.shuralyov.com/website/gido4ydmitshur opened an issue dmitri.shuralyov.com/go/generated: update implementation for 2020 spec changes4y