Add new baseline-diff command by autoantwort · Pull Request #1980 · microsoft/vcpkg-tool

autoantwort · 2026-04-08T09:40:55Z

If we use x-update-baseline we wanted to know which packages gets updated for our manifest file.

Copilot

Pull request overview

This PR introduces a new experimental x-baseline-diff command to help users understand which manifest dependencies would change when moving between two baselines (commit refs), integrating it into the vcpkg command dispatcher and localization system.

Changes:

Adds x-baseline-diff command implementation and registers it in command dispatch.
Extends configuration with a helper intended to detect whether the default registry is the built-in registry.
Adds new localized strings for command synopsis/output headings and “no change” output.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 12 comments.

Show a summary per file

File	Description
`src/vcpkg/configuration.cpp`	Adds a helper intended to detect built-in default registry kind (currently incorrect).
`include/vcpkg/configuration.h`	Exposes the new configuration helper in the public header.
`src/vcpkg/commands.cpp`	Wires the new command into the command table (currently has a syntax error in the include).
`src/vcpkg/commands.baseline-diff.cpp`	Implements the baseline comparison by generating two install plans and diffing versions.
`include/vcpkg/commands.baseline-diff.h`	Declares the new command entrypoint/metadata.
`locales/messages.json`	Adds user-facing strings for the command and output labels (some grammar issues).
`include/vcpkg/base/message-data.inc.h`	Declares new messages for localization enforcement (some grammar issues).

If we use x-update-baseline we wanted to know which packages gets updated for our manifest file.

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.

Billy O'Neal (BillyONeal) · 2026-04-15T00:34:20Z

+        return false;
+    }
+
+    void check_for_valid_sha(StringView sha)


Can this use check_commit_exists (from git.h) instead?

I see we can't do that in the general case because this kinda just blindly overwrites the baseline with a SHA and we don't necessarily have a git repo present for that.

However that makes this kinda broken because filesystem registries' baselines are not SHAs. We should only be checking that the input is a SHA when we know that registry has that constraint.

The code is registry.h/cpp could really benefit from some comments. What is a LockFile and LockFile::lockdata maps what to what?

Billy O'Neal (BillyONeal)

I think the general idea of such a command makes sense.

Most of what is here is nitpicks but the one about 'why manifest only' and 'what about non-git registries' are kinda foundational.

Should we have output like this printed by x-update-baseline instead of/in addition to what you've done here?

This needs at least one end to end test.

Thanks for the new command submission!

Billy O'Neal (BillyONeal) · 2026-04-15T00:45:30Z

+        msgCmdBaselineDiffSynopsis,
+        {"vcpkg x-baseline-diff <old_commit_sha> <newer_commit_sha>",
+         "vcpkg x-baseline-diff $(git rev-parse 2026.02.27) $(git rev-parse 2026.03.18)"},
+        Undocumented,


I think we need a docs page for this.

MicrosoftDocs/vcpkg-docs#584

Billy O'Neal (BillyONeal) · 2026-04-15T00:50:27Z

+                                        Triplet host_triplet)
+    {
+        const auto* manifest = paths.get_manifest();
+        if (manifest == nullptr)


I'm not sure I agree with this design: Because the user is explicitly passing in multiple baseline SHAs, it seems to really be more of a classic mode command as other registries aren't accounted for.

It seems like what you actually wanted is effectively this same diff but before/after what x-update-baseline does?

If this stays manifest only, it seems like one of the SHAs should be the one already in the manifest?

It seems like what you actually wanted is effectively this same diff but before/after what x-update-baseline does?

Yes. I wanted to know what x-update-baseline does mean to me in terms of port versions, or what it would mean.

If this stays manifest only, it seems like one of the SHAs should be the one already in the manifest?

At least if there is no second one. You are right, this would be more developer friendly.

Billy O'Neal (BillyONeal) · 2026-04-15T00:51:47Z

+        return false;
+    }
+
+    void check_for_valid_sha(StringView sha)


I see we can't do that in the general case because this kinda just blindly overwrites the baseline with a SHA and we don't necessarily have a git repo present for that.

However that makes this kinda broken because filesystem registries' baselines are not SHAs. We should only be checking that the input is a SHA when we know that registry has that constraint.

Billy O'Neal (BillyONeal)

(paraphrased by Billy O'Neal (@BillyONeal) )
Victor Romero (@vicroms): Anything that makes baselines more helpful and less obscure is a good thing
Augustin Popa (@AugP): This concept sounds like a good idea and I have no objections to it

Apply the print_lines nitpicks.

Co-authored-by: Billy O'Neal <bion@microsoft.com>

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 6 comments.

Victor Romero (vicroms) · 2026-05-28T19:26:38Z

+DECLARE_MESSAGE(BaselineDiffNoChange,
+                (),
+                "commit is a git commit here",
+                "No package that would have been installed was updated between the two commits")


Suggested change

"No package that would have been installed was updated between the two commits")

"No changes to installable packages between baselines")

I'm not sure if installable conveys the meaning but I'd prefer a terse description.

Maybe installed? Because the packages listed in the vcpkg.json are normally installed.

mostly written by claude code.

Copilot

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 5 comments.

Copilot

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 4 comments.

Copilot

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 5 comments.

…ry because there is also a builtin git registry which is not meant by is_default_builtin_registry

Billy O'Neal (BillyONeal) · 2026-07-06T22:29:49Z

+        if (!local_git_root.empty())
+        {
+            const auto* git_exe = paths.get_tool_path(console_diagnostic_context, Tools::GIT);
+            if (!git_exe) Checks::exit_fail(VCPKG_LINE_INFO);
+
+            GitRepoLocator locator{GitRepoLocatorKind::CurrentDirectory, local_git_root};
+            const auto resolve_baseline_ref = [&](StringView baseline) {
+                if (is_git_sha(baseline))
+                {
+                    return baseline.to_string();
+                }
+                auto maybe_resolved = git_resolve_to_full_sha(console_diagnostic_context, *git_exe, locator, baseline);
+                Checks::msg_check_exit(
+                    VCPKG_LINE_INFO, maybe_resolved.has_value(), msgInvalidGitRef, msg::value = baseline);
+                return *maybe_resolved.get();
+            };
+            old_baseline = resolve_baseline_ref(old_baseline_ref);
+            new_baseline = resolve_baseline_ref(new_baseline_ref);
+        }


I don't understand why we are trying to look at a local git root when this is a manifest mode command. What the local git situation is is totally irrelevant to manifest mode.

Billy O'Neal (BillyONeal) · 2026-07-06T22:33:22Z

It seems like what you actually wanted is effectively this same diff but before/after what x-update-baseline does?

Yes. I wanted to know what x-update-baseline does mean to me in terms of port versions, or what it would mean.

The problem with this is that x-update-baseline will update all registries in the manifest and this only cares about the default one. Only talking about the default registry means we could be showing an inaccurate view of what changes.

Inside x-update-baseline, we have this data for every registry we are updating and can show you the diff between the old world order and the new world order. Since the thing you seem to be actually trying to get out of this is what that diff looks like, should we fix x-update-baseline to print that rather than trying to add a new difficult-to-specify command like this?

I think this kind of "compare these two SHAs" should be a classic mode universe thing (which is exactly what portsdiff already intends to do) and the kind of comparison you're trying to do here would be better to compare 2 different manifests instead.

That said, I have some sympathy for "99% of people only use one registry so maybe this does what they actually want most of the time" but it sounds like even you really want this from x-update-baseline instead?

This is a psuedo-alternative to #1980 or at least to get similar results as intended there. I used GPT 5.5 in authoring this change. ```console PS C:\Dev\vcpkg-tool\src> C:\Dev\vcpkg-tool\out\build\Win-x64-Debug-WithArtifacts\vcpkg.exe x-update-baseline --dry-run --vcpkg-root C:\Dev\vcpkg Fetching registry information from https://github.com/microsoft/vcpkg (HEAD)... updated registry 'https://github.com/microsoft/vcpkg': e03dc9b29710050cd1018bc5674688108658d327 -> 42e4e33e1505c9f47b58c21e0f557c1571b751ee Updating baselines has resulted in the following version updates: Direct dependencies: curl: 8.20.0#1 -> 8.21.0#1 Transitive dependencies: zlib: 1.3.2 -> 1.3.2#1 ```

Billy O'Neal (BillyONeal) · 2026-08-05T19:41:50Z

I believe I implemented the outcome you were trying to get to here in #2057

Copilot AI review requested due to automatic review settings April 8, 2026 09:40

Copilot started reviewing on behalf of autoantwort April 8, 2026 09:41 View session

autoantwort force-pushed the feature/baseline-diff branch 2 times, most recently from 04a5e7a to 9d4ae6d Compare April 8, 2026 09:47

Copilot AI reviewed Apr 8, 2026

View reviewed changes

autoantwort force-pushed the feature/baseline-diff branch from 9d4ae6d to a9849db Compare April 8, 2026 09:54

Add new baseline-diff command

4a51845

If we use x-update-baseline we wanted to know which packages gets updated for our manifest file.

Copilot AI review requested due to automatic review settings April 8, 2026 10:14

autoantwort force-pushed the feature/baseline-diff branch from a9849db to 4a51845 Compare April 8, 2026 10:14

Copilot started reviewing on behalf of autoantwort April 8, 2026 10:15 View session

Copilot AI reviewed Apr 8, 2026

View reviewed changes

Comment thread src/vcpkg/commands.baseline-diff.cpp Outdated

Comment thread src/vcpkg/commands.baseline-diff.cpp

Comment thread src/vcpkg/commands.baseline-diff.cpp

Comment thread src/vcpkg/commands.baseline-diff.cpp

Billy O'Neal (BillyONeal) reviewed Apr 15, 2026

View reviewed changes

Billy O'Neal (BillyONeal) requested changes Apr 15, 2026

View reviewed changes

Apply the print_lines nitpicks.

93cc331

Billy O'Neal (BillyONeal) reviewed Apr 15, 2026

View reviewed changes

autoantwort and others added 2 commits April 23, 2026 17:25

Merge pull request #2 from BillyONeal/pr-1980-baseline-diff

650a931

Apply the print_lines nitpicks.

Apply suggestions from code review

1945ffd

Co-authored-by: Billy O'Neal <bion@microsoft.com>

Copilot AI review requested due to automatic review settings May 24, 2026 21:52

Copilot AI reviewed May 24, 2026

View reviewed changes

Victor Romero (vicroms) reviewed May 28, 2026

View reviewed changes

autoantwort added 2 commits June 8, 2026 14:24

More tests and edge case handling

5492f16

mostly written by claude code.

Merge branch 'main' into feature/baseline-diff

0bfc39b

Copilot AI review requested due to automatic review settings June 8, 2026 12:31

Copilot AI reviewed Jun 8, 2026

View reviewed changes

Comment thread src/vcpkg/commands.baseline-diff.cpp Outdated

Comment thread src/vcpkg/commands.baseline-diff.cpp

Comment thread src/vcpkg/commands.baseline-diff.cpp Outdated

Comment thread src/vcpkg/commands.baseline-diff.cpp

Comment thread locales/messages.json

fix msvc warning

a211bba

Copilot AI review requested due to automatic review settings June 14, 2026 12:23

Copilot AI reviewed Jun 14, 2026

View reviewed changes

Comment thread src/vcpkg/commands.baseline-diff.cpp Outdated

Comment thread locales/messages.json Outdated

Comment thread src/vcpkg/commands.baseline-diff.cpp Outdated

Comment thread azure-pipelines/end-to-end-tests-dir/baseline-diff.ps1

autoantwort added 3 commits June 14, 2026 14:32

update messages

12d0661

refs always have to be resolved

96ebbe9

update message

e7bcd2d

autoantwort added 2 commits June 14, 2026 15:30

better test

cdc0bef

add link to docs

f6943f7

Copilot AI review requested due to automatic review settings June 14, 2026 13:49

Copilot AI reviewed Jun 14, 2026

View reviewed changes

Comment thread src/vcpkg/commands.baseline-diff.cpp Outdated

Comment thread locales/messages.json Outdated

Comment thread locales/messages.json Outdated

Comment thread locales/messages.json

Comment thread locales/messages.json Outdated

autoantwort added 2 commits June 14, 2026 16:19

messages

0b38e43

Rename is_default_builtin_registry -> is_default_builtin_files_regist…

3e54317

…ry because there is also a builtin git registry which is not meant by is_default_builtin_registry

Billy O'Neal (BillyONeal) reviewed Jul 6, 2026

View reviewed changes

Billy O'Neal (BillyONeal) mentioned this pull request Jul 7, 2026

Add diff output to x-update-baseline #2057

Merged

	"No package that would have been installed was updated between the two commits")
	"No changes to installable packages between baselines")

Uh oh!

Conversation

autoantwort commented Apr 8, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Billy O'Neal (BillyONeal) left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Billy O'Neal (BillyONeal) left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!