web: non-blocking web_server, browser console mirroring, Tcl 9 fix by maliberty · Pull Request #10392 · The-OpenROAD-Project/OpenROAD

maliberty · 2026-05-11T06:42:13Z

Summary

Make web_server a transition point in scripts and a blocking suspend-the-prompt call at the interactive Tcl prompt, with the rest of the script running through the now-active server and browser-driven commands serializing against the main thread.

web_server Tcl proc: walks info frame for tclreadline::Loop to detect interactive vs. script context. Scripts return immediately so subsequent lines run as if typed in the browser console; interactive use calls web_server_wait_cmd which parks the main thread in Tcl_DoOneEvent so the launching terminal prompt is suppressed while the server is up. The event mask excludes TCL_FILE_EVENTS so tclreadline's stdin handler does not keep reading the terminal.
Tcl 9 cross-thread fix: TclEvaluator::eval marshals browser worker tcl_eval requests to the main thread via Tcl_ThreadQueueEvent + condition variable instead of calling Tcl_Eval directly on the worker. Tcl 9 isolates per-thread interpreter state, so a direct worker Tcl_Eval fails to see globals set on the main thread (e.g. set x 1 at the prompt followed by puts $x in the browser returns an unset-variable error). Evaluation uses Tcl_EvalEx(..., TCL_EVAL_GLOBAL) so commands run at the top-level namespace regardless of the Tcl call frame.
Browser console mirroring: a level-1 Tcl_CreateObjTrace broadcasts every main-thread top-level command as a console_input push frame, so script lines and terminal-typed commands appear in the browser Tcl console as >>> cmd. The trace filters by Tcl_GetCurrentThread() == mainThreadId() to avoid double-echoing browser-typed commands (the JS widget echoes them locally). Two Tcl_StackChannel transforms on stdout / stderr forward puts/error output through to the terminal and broadcast it to the browser as a log frame.
Browser-typed exit / web_server -stop wake the main thread via a no-op event queued with Tcl_ThreadQueueEvent + Tcl_ThreadAlert. The wait loop checks exitRequested() / stopRequested() and the calling site (Main.cc or web_server_wait_cmd) calls stop() and, on exit-request, std::exit(0) on the main thread.
Channel mirror passes the actual written count (not requested) to the broadcast so a partial write doesn't claim bytes that never reached the terminal.
WebServer::stop() integrates the new teardown (unstack channels, remove trace, restore exit, reset state) and remains callable from any thread via stopAndJoinIoThreads's self-join handling.
main.js renders console_input push frames as >>> cmd lines.

Tests adapted: TclEvaluator constructor now takes a shared std::mutex& and Tcl_ThreadId; tests pass nullptr for the thread id so the direct fast-path runs.

Type of Change

New feature

Verification

I have verified that the local build succeeds (./etc/Build.sh).
[ x I have run the relevant tests and they pass.
My code follows the repository's formatting guidelines.
I have included tests to prevent regressions.
I have signed my commits (DCO).

Make `web_server` a transition point in scripts and a blocking suspend-the-prompt call at the interactive Tcl prompt, with the rest of the script running through the now-active server and browser-driven commands serializing against the main thread. * `web_server` Tcl proc: walks `info frame` for `tclreadline::Loop` to detect interactive vs. script context. Scripts return immediately so subsequent lines run as if typed in the browser console; interactive use calls `web_server_wait_cmd` which parks the main thread in `Tcl_DoOneEvent` so the launching terminal prompt is suppressed while the server is up. The event mask excludes `TCL_FILE_EVENTS` so tclreadline's stdin handler does not keep reading the terminal. * Tcl 9 cross-thread fix: `TclEvaluator::eval` marshals browser worker `tcl_eval` requests to the main thread via `Tcl_ThreadQueueEvent` + condition variable instead of calling `Tcl_Eval` directly on the worker. Tcl 9 isolates per-thread interpreter state, so a direct worker `Tcl_Eval` fails to see globals set on the main thread (e.g. `set x 1` at the prompt followed by `puts $x` in the browser returns an unset-variable error). Evaluation uses `Tcl_EvalEx(..., TCL_EVAL_GLOBAL)` so commands run at the top-level namespace regardless of the Tcl call frame. * Browser console mirroring: a level-1 `Tcl_CreateObjTrace` broadcasts every main-thread top-level command as a `console_input` push frame, so script lines and terminal-typed commands appear in the browser Tcl console as `>>> cmd`. The trace filters by `Tcl_GetCurrentThread() == mainThreadId()` to avoid double-echoing browser-typed commands (the JS widget echoes them locally). Two `Tcl_StackChannel` transforms on stdout / stderr forward `puts`/error output through to the terminal and broadcast it to the browser as a `log` frame. * Browser-typed `exit` / `web_server -stop` wake the main thread via a no-op event queued with `Tcl_ThreadQueueEvent` + `Tcl_ThreadAlert`. The wait loop checks `exitRequested()` / `stopRequested()` and the calling site (`Main.cc` or `web_server_wait_cmd`) calls `stop()` and, on exit-request, `std::exit(0)` on the main thread. * Channel mirror passes the actual `written` count (not requested) to the broadcast so a partial write doesn't claim bytes that never reached the terminal. * `WebServer::stop()` integrates the new teardown (unstack channels, remove trace, restore `exit`, reset state) and remains callable from any thread via `stopAndJoinIoThreads`'s self-join handling. * main.js renders `console_input` push frames as `>>> cmd` lines. Tests adapted: `TclEvaluator` constructor now takes a shared `std::mutex&` and `Tcl_ThreadId`; tests pass `nullptr` for the thread id so the direct fast-path runs. Signed-off-by: Matt Liberty <mliberty@precisioninno.com>

github-actions · 2026-05-11T06:50:04Z

clang-tidy review says "All clean, LGTM! 👍"

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 07954badef

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-05-11T06:54:17Z

+  if { $in_readline } {
+    web::web_server_wait_cmd
+  }


Keep web_server blocking while scripts mutate shared state

Returning immediately in script mode means the main Tcl thread can keep running design-mutating commands while web worker threads continue servicing requests, but the worker-side serialization (tcl_eval_->mutex in handlers such as SelectHandler::handleSelect) does not protect against concurrent main-thread STA/DB access. This reintroduces unsynchronized cross-thread access to components explicitly marked non-thread-safe and can cause races or crashes when a script runs after web_server and a browser is connected.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-05-11T06:54:17Z

+  while (!exitRequested() && !stopRequested() && isRunning()) {
+    Tcl_DoOneEvent(kEventMask);


Stop spinning when Tcl_DoOneEvent reports no work

This loop ignores the return value of Tcl_DoOneEvent; when there are no eligible timer/idle/window handlers, Tcl can return 0 immediately instead of blocking, so the loop busy-spins until stopRequested/exitRequested flips. In headless web sessions this can peg a CPU core while waiting for user actions.

Useful? React with 👍 / 👎.

gemini-code-assist

Code Review

This pull request refactors the web server's integration with the Tcl event loop to support Tcl 9's thread isolation and improve the interactive experience. Key changes include implementing cross-thread marshalling for Tcl commands via Tcl_ThreadQueueEvent, mirroring Tcl's stdout and stderr channels to the browser console using Tcl_StackChannel, and broadcasting top-level main-thread commands to connected clients. The WebServer API was updated to replace the blocking waitForStop with a Tcl-native event loop in runEventLoopUntilStop. Feedback identified a signature mismatch in the Tcl channel output procedure where a parameter should be char* instead of const char* to comply with the Tcl API.

gemini-code-assist · 2026-05-11T06:55:59Z

+static int webChannelOutputProc(ClientData instanceData,
+                                const char* buf,
+                                int toWrite,
+                                int* errorCodePtr)
+{
+  auto* m = static_cast<WebChannelMirror*>(instanceData);
+
+  // Forward to the next channel in the stack so the terminal still
+  // sees the output.
+  const Tcl_ChannelType* parent_type = Tcl_GetChannelType(m->parent);
+  Tcl_DriverOutputProc* parent_output = Tcl_ChannelOutputProc(parent_type);
+  ClientData parent_data = Tcl_GetChannelInstanceData(m->parent);
+  const int written = parent_output(
+      parent_data, const_cast<char*>(buf), toWrite, errorCodePtr);


The signature of webChannelOutputProc does not match the Tcl_DriverOutputProc type expected by Tcl_ChannelType.outputProc. The buf parameter should be char*, not const char*.

This mismatch can cause compilation errors and relies on unsafe type conversions. Correcting the signature to char* buf aligns with the Tcl API, improves type safety, and removes the need for const_cast when calling the parent channel's output procedure.

Suggested change

static int webChannelOutputProc(ClientData instanceData,

const char* buf,

int toWrite,

int* errorCodePtr)

{

auto* m = static_cast<WebChannelMirror*>(instanceData);

// Forward to the next channel in the stack so the terminal still

// sees the output.

const Tcl_ChannelType* parent_type = Tcl_GetChannelType(m->parent);

Tcl_DriverOutputProc* parent_output = Tcl_ChannelOutputProc(parent_type);

ClientData parent_data = Tcl_GetChannelInstanceData(m->parent);

const int written = parent_output(

parent_data, const_cast<char*>(buf), toWrite, errorCodePtr);

static int webChannelOutputProc(ClientData instanceData,

char* buf,

int toWrite,

int* errorCodePtr)

{

auto* m = static_cast<WebChannelMirror*>(instanceData);

// Forward to the next channel in the stack so the terminal still

// sees the output.

const Tcl_ChannelType* parent_type = Tcl_GetChannelType(m->parent);

Tcl_DriverOutputProc* parent_output = Tcl_ChannelOutputProc(parent_type);

ClientData parent_data = Tcl_GetChannelInstanceData(m->parent);

const int written = parent_output(

parent_data, buf, toWrite, errorCodePtr);

References

API usability and compatibility with external driver signatures (like Tcl_DriverOutputProc) can be prioritized over strict const-correctness to avoid unsafe type conversions and ensure correct linkage.

maliberty self-assigned this May 11, 2026

github-actions Bot added the size/L label May 11, 2026

maliberty requested a review from gadfort May 11, 2026 06:43

chatgpt-codex-connector Bot reviewed May 11, 2026

View reviewed changes

gemini-code-assist Bot reviewed May 11, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

web: non-blocking web_server, browser console mirroring, Tcl 9 fix#10392

web: non-blocking web_server, browser console mirroring, Tcl 9 fix#10392
maliberty wants to merge 1 commit into
The-OpenROAD-Project:masterfrom
The-OpenROAD-Project-staging:web-non-blocking

maliberty commented May 11, 2026

Uh oh!

github-actions Bot commented May 11, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot May 11, 2026

Uh oh!

chatgpt-codex-connector Bot May 11, 2026

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot May 11, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

		while (!exitRequested() && !stopRequested() && isRunning()) {
		Tcl_DoOneEvent(kEventMask);

Conversation

maliberty commented May 11, 2026

Summary

Type of Change

Verification

Uh oh!

github-actions Bot commented May 11, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant