build: upgrade -dlgo version to Go 1.24.0

[levisyin]

I have checked it via go run build/ci.go sanitycheck

[MariusVanDerWijden]

Still waiting on BLST to cut a release before we can merge this

[holiman]

When doing the golang upgrade, it's probably better to keep all the changes together.
Go-ethereum retains compatibility with two golang versions, so updating to 1.24 means dropping support for 1.22.

See #30323 for an idea about what needs to be changed, in addition to the checksums (mostly CI workflows and readme).

[levisyin]

When doing the golang upgrade, it's probably better to keep all the changes together. Go-ethereum retains compatibility with two golang versions, so updating to 1.24 means dropping support for 1.22.

See #30323 for an idea about what needs to be changed, in addition to the checksums (mostly CI workflows and readme).

Okay, I will try to change all related things.

pr was merged

[MariusVanDerWijden]

We should also update golangci-lint to the newest version which supports 1.24: https://github.com/golangci/golangci-lint/releases/tag/v1.64.4

[levisyin]

We should also update golangci-lint to the newest version which supports 1.24: https://github.com/golangci/golangci-lint/releases/tag/v1.64.4

Okay, I will do this now

[rjl493456442]

Any particular reason to bump the minimal version requirement?

[levisyin]

See the comment by @holiman

Go-ethereum retains compatibility with two golang versions, so updating to 1.24 means dropping support for 1.22

So I updated the minimal version requirement.

[levisyin]

WARN The linter 'tenv' is deprecated (since v1.64.0) due to: Duplicate feature another linter. Replaced by usetesting.

@MariusVanDerWijden Hi sir, shall we open usetesting linter? As this linter will report issues:

cmd/clef/run_test.go:60:15: os.MkdirTemp() could be replaced by t.TempDir() in runClef (usetesting)
        ddir, err := os.MkdirTemp("", "cleftest-*")
                     ^
cmd/utils/history_test.go:90:14: os.MkdirTemp() could be replaced by t.TempDir() in TestHistoryImportAndExport (usetesting)
        dir, err := os.MkdirTemp("", "history-export-test")
                    ^
core/txpool/blobpool/blobpool_test.go:455:16: os.MkdirTemp() could be replaced by t.TempDir() in TestOpenDrops (usetesting)
        storage, _ := os.MkdirTemp("", "blobpool-")
                      ^
core/txpool/blobpool/blobpool_test.go:778:16: os.MkdirTemp() could be replaced by t.TempDir() in TestOpenIndex (usetesting)
        storage, _ := os.MkdirTemp("", "blobpool-")
                      ^
core/txpool/blobpool/blobpool_test.go:867:16: os.MkdirTemp() could be replaced by t.TempDir() in TestOpenHeap (usetesting)
        storage, _ := os.MkdirTemp("", "blobpool-")
                      ^
core/txpool/blobpool/blobpool_test.go:954:16: os.MkdirTemp() could be replaced by t.TempDir() in TestOpenCap (usetesting)
        storage, _ := os.MkdirTemp("", "blobpool-")
                      ^
core/txpool/blobpool/blobpool_test.go:1044:16: os.MkdirTemp() could be replaced by t.TempDir() in TestChangingSlotterSize (usetesting)
        storage, _ := os.MkdirTemp("", "blobpool-")
                      ^
core/txpool/blobpool/blobpool_test.go:1511:17: os.MkdirTemp() could be replaced by t.TempDir() in TestAdd (usetesting)
                storage, _ := os.MkdirTemp("", "blobpool-")
                              ^
crypto/crypto_test.go:184:13: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestLoadECDSA (usetesting)
                f, err := os.CreateTemp("", "loadecdsa_test.*.txt")
                          ^
crypto/crypto_test.go:205:12: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestSaveECDSA (usetesting)
        f, err := os.CreateTemp("", "saveecdsa_test.*.txt")
                  ^
crypto/signify/signify_test.go:36:18: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestSignify (usetesting)
        tmpFile, err := os.CreateTemp("", "")
                        ^
crypto/signify/signify_test.go:78:18: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestSignifyTrustedCommentTooManyLines (usetesting)
        tmpFile, err := os.CreateTemp("", "")
                        ^
crypto/signify/signify_test.go:101:18: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestSignifyTrustedCommentTooManyLinesLF (usetesting)
        tmpFile, err := os.CreateTemp("", "")
                        ^
crypto/signify/signify_test.go:124:18: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestSignifyTrustedCommentEmpty (usetesting)
        tmpFile, err := os.CreateTemp("", "")
                        ^
internal/era/era_test.go:40:12: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestEra1Builder (usetesting)
        f, err := os.CreateTemp("", "era1-test")
                  ^
node/config_test.go:56:15: os.CreateTemp("", ...) could be replaced by os.CreateTemp(t.TempDir(), ...) in TestDatadirCreation (usetesting)
        file, err := os.CreateTemp("", "")

[MariusVanDerWijden]

I think we should do that in a follow-up pr, same with bumping the min-version imo

[levisyin]

I think we should do that in a follow-up pr, same with bumping the min-version imo

@MariusVanDerWijden Done.

I have reverted the min go version requirement in README.md, and brought tenv linter back.

[holiman]

same with bumping the min-version imo

I figure that the text in the readme is an instruction / promise to the uesrs that geth work with the version posted there. Since this PR removes the CI-workflows which detects when we break that compatibility, it's overall a safer bet to change it to 1.23.

I mean, if we do support 1.22 a while longer, despite not explicitly promising to do so, nobody will complain. Whereas if some PR breaks it without CI detecting it, somebody is bound to complain.

But yeah, it's up to you (Marius et al)!

[levisyin]

@MariusVanDerWijden Hi sir, Is there anything else I need to update here?

[fjl]

Not really. We will merge this when we feel ready for the upgrade to Go 1.24.

[fjl]

This is not correct. We always want to keep a builder around for the previous Go releases, to catch regressions where we accidentally start using Go 1.24 features.