vendor: update docker/docker and docker/cli to v26.0.0

[thaJeztah]

• supersedes / closes build(deps): bump github.com/docker/docker from 26.0.0-rc1+incompatible to 26.0.0-rc3+incompatible #4786

vendor: golang.org/x/sys v0.18.0

full diff: golang/sys@v0.16.0...v0.18.0

vendor: github.com/docker/docker v26.0.0

full diff: moby/moby@v26.0.0-rc1...v26.0.0

vendor: github.com/docker/cli v26.0.0

no changes in vendored code

full diff: docker/cli@v26.0.0-rc1...v26.0.0

[thaJeztah]

Arf, the go:version hits again;

0.175 + xx-go build '-gcflags=' -ldflags '-X github.com/moby/buildkit/version.Version=b53a13e -X github.com/moby/buildkit/version.Revision=b53a13e4f5c8d7e82716615e0f23656893df89af -X github.com/moby/buildkit/version.Package=github.com/moby/buildkit -extldflags '"'"'-static'"'" -tags 'osusergo netgo static_build seccomp ' -o /usr/bin/buildkitd ./cmd/buildkitd
181.8 # github.com/docker/docker/libnetwork/internal/resolvconf
181.8 vendor/github.com/docker/docker/libnetwork/internal/resolvconf/resolvconf.go:509:2: clear requires go1.21 or later (-lang was set to go1.16; check go.mod)

[thaJeztah]

• opened resolvconf: add //go:build directives to prevent downgrading to go1.16 language moby#47578

[tonistiigi]

@thaJeztah Could you update this to GA version

[thaJeztah]

Hm, some failures like this, which seems related. Wondering if the new implementation has a stable output (in which case, we can update the tests), or if the order of items in the resolv.conf should be considered "unstable", and rewrite the tests to account for that; @robmry could you have a peek?

time="2024-04-09T09:03:46Z" level=info msg="No non-localhost DNS nameservers are left in resolv.conf. Using default external servers"
    resolvconf_test.go:131: 
        	Error Trace:	/src/executor/oci/resolvconf_test.go:131
        	Error:      	Not equal: 
        	            	expected: "\noptions ndots:0\nnameserver 8.8.8.8\nnameserver 8.8.4.4\nnameserver 2001:4860:4860::8888\nnameserver 2001:4860:4860::8844"
        	            	actual  : "nameserver 8.8.8.8\nnameserver 8.8.4.4\nnameserver 2001:4860:4860::8888\nnameserver 2001:4860:4860::8844\noptions ndots:0\n"
        	            	
        	            	Diff:
        	            	--- Expected
        	            	+++ Actual
        	            	@@ -1,3 +1 @@
        	            	-
        	            	-options ndots:0
        	            	 nameserver 8.8.8.8
        	            	@@ -6 +4,3 @@
        	            	 nameserver 2001:4860:4860::8844
        	            	+options ndots:0
        	            	+
        	Test:       	TestResolvConf/TestRegenerateResolvconfToAddLocalDNS

Old

(note: newline at start, no newline at end in expected: "\noptions ndots:0\nnameserver 8.8.8.8\nnameserver 8.8.4.4\nnameserver 2001:4860:4860::8888\nnameserver 2001:4860:4860::8844":

options ndots:0
nameserver 8.8.8.8
nameserver 8.8.4.4
nameserver 2001:4860:4860::8888
nameserver 2001:4860:4860::8844

New

(note: no newline at start, but newline at end in actual : "nameserver 8.8.8.8\nnameserver 8.8.4.4\nnameserver 2001:4860:4860::8888\nnameserver 2001:4860:4860::8844\noptions ndots:0\n")

nameserver 8.8.8.8
nameserver 8.8.4.4
nameserver 2001:4860:4860::8888
nameserver 2001:4860:4860::8844
options ndots:0

[robmry]

Hm, some failures like this, which seems related. Wondering if the new implementation has a stable output (in which case, we can update the tests)

Yes, the new output is stable ... before the 'resolvconf' changes, the generated file was effectively a sed'd version of the host's.

Now, the host's input file is read into an internal structure, and the output is generated based on a text-template - so it'll always look the same, even if the shape of the input changes.

[thaJeztah]

Updated, and adjusted the test

@crazy-max @tonistiigi @robmry PTAL

[crazy-max]

LGTM