Improved details for build and pull by craigloewen-msft · Pull Request #40596 · microsoft/WSL

craigloewen-msft · 2026-05-19T18:13:19Z

Summary of the Pull Request

Added some small details to wslc build and wslc pull.

Specifically this just adds a 'total MB' to build and progress bars and total layer sizes to pull.

Pull

Before:

After:

Build

Before:

After:

PR Checklist

Closes: Link to issue #xxx
Communication: I've discussed this with core contributors already. If work hasn't been agreed, this work might be rejected
Tests: Added/updated if needed and all pass
Localization: All end user facing strings can be localized
Dev docs: Added/updated if needed
Documentation updated: If checked, please file a pull request on our docs repo and link it here: #xxx

Detailed Description of the Pull Request / Additional comments

Scoped it to be as small as possible.

Validation Steps Performed

Tested behaviour before and after.

Copilot

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Adds richer progress reporting to wslc build and wslc pull: pull now shows progress bars with current/total bytes, and build shows per-layer download progress lines that update in place along with colored build-step output.

Changes:

Plumb current/total byte counts from BuildKit status through BuildKitStatus JSON, reportProgress, and into the callbacks.
Render a fixed-width progress bar with FormatBytes output in ImageProgressCallback, and truncate to console width to avoid wrap artifacts.
In BuildImageCallback, maintain a per-entry m_pullLines map rendered each redraw, and colorize permanent build-step lines in teal.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
src/windows/wslcsession/WSLCSession.cpp	Extends `reportProgress` to forward current/total and emits per-entry progress when `entry.total > 0`.
src/windows/wslc/services/ImageProgressCallback.cpp	Replaces percent text with a 30-char progress bar plus formatted byte counts; truncates to console width.
src/windows/wslc/services/BuildImageCallback.h	Adds `m_pullLines` map and `<map>` include.
src/windows/wslc/services/BuildImageCallback.cpp	Stores per-id pull status, redraws periodically, and wraps permanent lines in teal ANSI escapes.
src/windows/inc/docker_schema.h	Adds `current`/`total` fields to `BuildKitStatus` deserialization.
src/shared/inc/stringshared.h	New `FormatBytes` helper producing GB/MB/KB/B strings.

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 5 comments.

Copilot

Pull request overview

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

benhillis · 2026-05-20T15:36:02Z

    // accepting any non-empty id, so future or unrelated id usage defaults to permanent.
    const bool isLog = (id != nullptr && std::string_view{id} == "log");

+    // Pull/download progress: update the per-entry map so Redraw can show each entry


I'm not sure we are handling this correctly when m_isConsole is false. We should probably skip the whole callback stuff when non in an interactive terminal.

I think this would break AIs who are scraping wslc.exe build output. Right now we output something like this:

PS C:\cDev\copilot\build-test> cat .\dockerfilesimple.log Building image from directory: . [1/2] FROM docker.io/library/ubuntu:24.04@sha256:c4a8d5503dfb2a3eb8ab5f807da5bc69a85730fb49b5cfca2330194ebcc41c7b [2/2] RUN echo 123 exporting to image | exporting layers | writing image sha256:780c43a5ab95735b080898e63db96c3feebdd05591a0a934bc3252fe70201004 | naming to docker.io/library/test PS C:\cDev\copilot\build-test>

I think this is pretty reasonable and gives the right info on all the steps.

…m/microsoft/WSL into user/crloewen/better-build-output

kvega005 · 2026-05-20T21:06:08Z

+inline std::wstring FormatBytes(uint64_t bytes)
+{


I am wondering if we could move this to wsl::windows::common::string and use the windows StrFormatByteSizeEx. It does have a slightly different oputput. StrFormatByteSizeEx uses 1024-based math with the Windows-conventional labels (KB, MB, GB, TB) — i.e. 1.00 KB means 1024 bytes. The current helper is 1000-based. So switching will change the numbers users see during pulls which would diverge from docker's behavior. However, it does give localized decimal seperator. I could see this going either way.

ALSO, if we are making a helper for this, we should probably get rid of the other FormatBytes method in ContainerTasks.cpp:37 and just use a common helper.

Opportunity to centralize this, such as moving the one from ContainerTasks into stringshared.h

I'm not sure if Craig's PR is the right place to do that consolidation, but at the very least using the same method for it is appropriate.

kvega005 · 2026-05-20T21:17:02Z

+        while (!wide.empty() && (wide.back() == L'\n' || wide.back() == L'\r'))
+        {
+            terminator.insert(terminator.begin(), wide.back());
+            wide.pop_back();
+        }


This is a bit hard to reason about. I think this can be simplified with std::wstring::find_last_not_of(L"\r\n").

const auto bodyLength = wide.find_last_not_of(L"\r\n") + 1; // or textLength or whatever you prefer (I am bad at naming stuff) const auto newlines = wide.substr(bodyLength); wide.resize(bodyLength); WriteTerminal(std::format(L"\033[36m{}\033[0m{}", wide, newlines));

kvega005 · 2026-05-20T21:37:11Z

    const bool isLog = (id != nullptr && std::string_view{id} == "log");

+    // Pull/download progress: update the per-entry map so Redraw can show each entry
+    // on a single line that updates in place.


Nit: The OnProgress body asks the same three questions in two different places — "does the message have an id?", "is it the log sentinel?", "is it per-layer pull progress?" — and the id != nullptr && *id != '\0' check appears several times. The verbose/non-TTY branch and the interactive branch also independently re-derive what kind of message this is.

const std::string_view idView = (id != nullptr) ? id : std::string_view{}; const bool isLog = (idView == "log"); const bool isPullProgress = (!idView.empty() && total > 0 && !isLog); if (m_verbose || !m_isConsole) { if (!isPullProgress) { wprintf(L"%hs", status); } return S_OK; } if (isPullProgress) { return HandlePullProgress(idStr, status); } if (isLog) { return HandleLogChunk(status); } return HandleStepHeader(status);

kvega005 · 2026-05-20T21:55:49Z

+            terminator.insert(terminator.begin(), wide.back());
+            wide.pop_back();
+        }
+        wide = std::format(L"\033[36m{}\033[0m{}", wide, terminator);


Really nit, don't hate me for this. There are a lot of these ANSI escape sequences scattered throught the file, it would be nice to have to have these defined as constants for readability. Either at the top of this file or a common header. (IDK if its used elsewhere).

For example, this format could read like: wide = std::format(L"{}{}{}{}", ansi::Cyan, wide, ansi::ResetColor, terminator);
Then its a lot easier to understand what is happening.

Actually, I did a brief search with copilot, and alledgedly its also used in WSLCTests.cpp, WSLCE2EHelpers.h, ImageProgressCallback.cpp, BuildImageCallback.cpp. So this could warrant placing in stringshared.h or deps.h. The test files i mentioned use the VT_* constants.

I was about to make the same comment. I'm not sure this PR should do a lot of refactoring, which could make it a lot bigger than it should be, but at the very least define the VT sequences as constants so the construction is a lot easier to read and so we can do that refactoring more easily later.

kvega005 · 2026-05-20T22:03:56Z

+    // Per-entry pull progress lines, keyed by entry id. Updated in place by Redraw.
+    std::map<std::string, std::string> m_pullLines;


The comment doesn't capture the load-bearing detail here: std::map's ordered iteration is what makes the visual "in place" rendering work. Redraw walks the map in key order each frame, so each layer id consistently lands on the same row — swap this for unordered_map and entries would shuffle between frames. Worth calling that out.

kvega005 · 2026-05-20T22:09:25Z

+        std::wstring bar(c_progressBarWidth, L' ');
+        for (int i = 0; i < filled; ++i)
+        {
+            bar[i] = L'=';
+        }
+
+        if (filled < c_progressBarWidth)
+        {
+            bar[filled] = L'>';
+        }


This pre-fills with spaces and then overwrites characters by index which is fine, but std::wstring already has constructors and append(n, ch) that express this directly without an explicit loop or index arithmetic. The intent reads more clearly when the bar is constructed segment-by-segment:

std::wstring bar; bar.reserve(c_progressBarWidth); bar.append(filled, L'='); bar.append(L">"); bar.resize(c_progressBarWidth, L' ');

Recommend to use a std::wostringstream for these constructions rather than doing repeated manipulation of the same string. The stringstreams were designed to do that efficiently. Then convert once to a wstring when all manipulations are complete.

kvega005 · 2026-05-20T22:15:40Z

+    while (line.size() < static_cast<size_t>(visibleWidth))
    {
        line += L' ';
    }


nit: line.resize(visibleWidth, L' '); - no need for while loop

dkbennett

Echo a lot of the great feedback Kevin had. I like that this PR is relatively small though, and if refactoring things would lead to it growing significantly I would recommend not doing that and having a separate PR for that refactoring. But definitely all for things that make the refactoring easier at the very least.

dkbennett · 2026-05-20T22:14:29Z

+inline std::wstring FormatBytes(uint64_t bytes)
+{


Opportunity to centralize this, such as moving the one from ContainerTasks into stringshared.h

I'm not sure if Craig's PR is the right place to do that consolidation, but at the very least using the same method for it is appropriate.

dkbennett · 2026-05-20T22:28:08Z

+            terminator.insert(terminator.begin(), wide.back());
+            wide.pop_back();
+        }
+        wide = std::format(L"\033[36m{}\033[0m{}", wide, terminator);


I was about to make the same comment. I'm not sure this PR should do a lot of refactoring, which could make it a lot bigger than it should be, but at the very least define the VT sequences as constants so the construction is a lot easier to read and so we can do that refactoring more easily later.

kvega005

On a successful build, the dim scrolling window gets erased by the final CollapseWindow(), so once the build finishes only the teal step headers ([1/4] FROM …, [2/4] RUN …) remain in scrollback. The replay branch in the destructor only fires when std::uncaught_exceptions() indicates failure.

For comparison, DOCKER_BUILDKIT=0 docker build keeps all RUN stdout in scrollback interleaved with step headers. So things like npm WARN …, pip resolver notes, or a BUILD SUCCESSFUL in 47s line from a tool inside the container aren't recoverable after a successful wslc build without re-running it.

Not sure if that's intentional (the compact view is genuinely nice) — flagging in case it isn't. A couple of options if you do want to address it:

Drop the std::uncaught_exceptions() check so the log replays on success too. m_allLines is already populated and bounded by c_maxAllLinesBytes.
Or add a --progress=plain / --quiet-style opt-out (could just reuse the existing m_verbose path).

craigloewen-msft added 2 commits May 19, 2026 14:05

Improved details for build and pull

050e605

Updated clang format

82bf7ec

Copilot AI review requested due to automatic review settings May 19, 2026 18:13

craigloewen-msft requested a review from a team as a code owner May 19, 2026 18:13

Copilot AI reviewed May 19, 2026

View reviewed changes

Copilot started reviewing on behalf of craigloewen-msft May 19, 2026 18:38 View session

Potential fix for pull request finding

40fa39a

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings May 19, 2026 18:40

Copilot AI reviewed May 19, 2026

View reviewed changes

Copilot started reviewing on behalf of craigloewen-msft May 19, 2026 18:57 View session

craigloewen-msft added 2 commits May 19, 2026 15:01

PR fixes

f218d7b

PR fixes

0ac7f58

benhillis requested a review from Copilot May 19, 2026 19:54

Copilot AI reviewed May 19, 2026

View reviewed changes

Comment thread src/windows/wslcsession/WSLCSession.cpp Outdated

Comment thread src/windows/wslc/services/BuildImageCallback.cpp Outdated

Comment thread src/shared/inc/stringshared.h Outdated

Comment thread src/windows/wslc/services/ImageProgressCallback.cpp Outdated

Copilot started reviewing on behalf of benhillis May 19, 2026 20:08 View session

benhillis reviewed May 20, 2026

View reviewed changes

Comment thread src/windows/wslc/services/BuildImageCallback.cpp

benhillis reviewed May 20, 2026

View reviewed changes

craigloewen-msft added 2 commits May 20, 2026 14:08

Merge branch 'user/crloewen/better-build-output' of https://github.co…

5af285d

…m/microsoft/WSL into user/crloewen/better-build-output

Added in PR fixes

3983eb8

kvega005 self-requested a review May 20, 2026 20:49

kvega005 reviewed May 20, 2026

View reviewed changes

dkbennett reviewed May 20, 2026

View reviewed changes

kvega005 reviewed May 20, 2026

View reviewed changes

		// Per-entry pull progress lines, keyed by entry id. Updated in place by Redraw.
		std::map<std::string, std::string> m_pullLines;

Conversation

craigloewen-msft commented May 19, 2026

Summary of the Pull Request

Pull

Build

PR Checklist

Detailed Description of the Pull Request / Additional comments

Validation Steps Performed

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!

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!

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!

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!

kvega005 May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kvega005 May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kvega005 May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

kvega005 May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kvega005 May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dkbennett left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kvega005 May 20, 2026 •

edited

Loading

kvega005 May 20, 2026 •

edited

Loading

kvega005 May 20, 2026 •

edited

Loading

kvega005 May 20, 2026 •

edited

Loading

kvega005 May 20, 2026 •

edited

Loading