如何检测Tcl线程fileevent或脚本之后的错误

Question

在我的主线程中，我试图检测并响应工作线程的fileevent脚本中发生的错误。我编写了一个测试脚本，试图在四种情况下检测线程错误：

1. 同步线程::发送，线程中的立即错误
2. 同步线程::发送，脚本后的线程错误
3. 异步线程::发送，线程中的立即错误
4. 异步线程::发送，脚本后的线程错误

我在测试线程中使用了一个after命令来（希望）模仿一个fileevent脚本。正如预期的那样，通过catch :: send命令的catch捕获case 1线程错误，并且thread :: errorproc过程和我的自定义bgerror过程捕获了case 3线程错误。案例2和4线程错误反映在主线程中，但我似乎无法通过我的自定义过程捕获它们。

以下是我的测试脚本：

    #!/usr/bin/env tclsh


    package require Thread;


    # Custom error handling procedure for errors during asynchronous thread::send
    proc error_proc {id error} {
            puts "\n---------- error_proc: $id $error **********\n";
    }
    thread::errorproc error_proc;


    # Custom error and bgerror procedures to try to see the thread error
    rename error original_error;
    proc error {err} {
            puts "---------- error: $err\n";
    }
    proc bgerror {err} {
            puts "---------- bgerror: $err\n";
    }


    # Non-blocking wait procedure to keep the output in order
    proc waitms {ms} {
            set ::done 0;
            after $ms {set ::done 1};
            vwait ::done;
    }


    # Create the worker thread
    set tid [thread::create {

            proc test_error_immediately {s} {
                    error "test_error_immediately ($s)";
            }

            proc test_error_after {ms s} {
                    after $ms [list error "test_error_after ($s)"];
            }

            thread::wait;

            puts "---------- Exiting thread";
    }]


    # Test case 1 - synchronous thread::send, immediate error
    if {[catch {thread::send $tid [list test_error_immediately "Case 1"]} result]} {
            set state "ERROR";
    } else {
            set state "OK";
    }
    puts "---------- Synchronous test_error_immediately $state ($result)\n"


    # Test case 2 - synchronous thread::send, error in thread after script
    if {[catch {thread::send $tid [list test_error_after 1000 "Case 2"]} result]} {
            set state "ERROR";
    } else {
            set state "OK";
    }
    puts "---------- Synchronous test_error_after $state ($result)\n";


    # Wait so that the output is in order
    waitms 1500;


    # Test case 3 - asynchronous thread::send, immediate error
    if {[catch {thread::send -async $tid [list test_error_immediately "Case 3"] result} catch_result]} {
            set state "ERROR";
            set result $catch_result;
    } else {
            set state "OK";
            vwait result;
    }
    puts "---------- Asynchronous test_error_immediately $state ($result)\n";


    # Wait so that the output is in order
    waitms 100;


    # Test case 4 - asynchronous thread::send, error in thread after script
    if {[catch {thread::send -async $tid [list test_error_after 1000 "Case 4"] result} catch_result]} {
            set state "ERROR";
            set result $catch_result;
    } else {
            set state "OK";
            vwait result;
    }
    puts "---------- Asynchronous test_error_after $state ($result)\n";


    # Clean up
    if {[thread::exists $tid]} {

            after 2000 {
                    thread::release $::tid;
                    puts "\n---------- $::tid released\n";
            }
    }


    puts "---------- Waiting forever ...\n";


    vwait forever;

以下是我电脑上的输出：

    ---------- Synchronous test_error_immediately ERROR (test_error_immediately (Case 1))

    ---------- Synchronous test_error_after OK (after#0)

    test_error_after (Case 2)
        while executing
    "error {test_error_after (Case 2)}"
        ("after" script)

    ---------- error_proc: tid0x7f5673e9d700 test_error_immediately (Case 3)
        while executing
    "error "test_error_immediately ($s)""
        (procedure "test_error_immediately" line 2)
        invoked from within
    "test_error_immediately {Case 3}" **********

    ---------- Asynchronous test_error_immediately OK (test_error_immediately (Case 3))

    ---------- bgerror: test_error_immediately (Case 3)

    ---------- Asynchronous test_error_after OK (after#1)

    ---------- Waiting forever ...

    test_error_after (Case 4)
        while executing
    "error {test_error_after (Case 4)}"
        ("after" script)

    ---------- tid0x7f5673e9d700 released

    ---------- Exiting thread

我正在使用Tcl 8.5和Ubuntu 10.04 LTS。

对于如何检测线程的fileevent或脚本之后发生的错误，您有任何其他建议吗？

的问候，
smercer