Bootstrap Tcl track

[sshine]

• Update docs: Initial docs #4
• Implement Hello World: hello-world: Create exercise #5
• CI for checking validity of example
• Set up commit hook and policy for PR-driven development
• Find json library for extracting canonical data for exercises
• Build test generator

[sshine]

Going over the checklist:

• Select programming language variant (not applicable)
• Select testing framework: tcltest ships with tclsh8.4+
• Write first exercise: hello-world: Create exercise #5 (hello-world)
• Plan initial core exercises: The plan is to first copy the Ruby core track.
• Add the track to the site as inactive
• Request a track icon: Icon for Tcl language track website-icons#26
• Plan optional exercises
• Write "getting started" instructions: Installation instructions not done yet; work in progress: Initial docs #4.

[sshine]

https://wuhrr.wordpress.com/2011/03/28/getting-started-with-tcltest/

[sshine]

Idea by @sebres for making single-file testing fail fast:

proc test args {
    upvar ::tcltest::numTests numTests
    set numTestsFailed $numTests(Failed)
    uplevel [list ::tcltest::test {*}$args]
    if {$numTests(Failed) > $numTestsFailed} {
        puts "*** EXIT ON ERROR ***"
        exit 1
    }
}

Updated: Fixed nt -> numTests in all places. :-)

[sebres]

well, initially I wrote:

-    upvar ::tcltest::numTests numTests
+    upvar ::tcltest::numTests nt

so either rename nt var everywhere (or apply fix above;)

[sshine]

Thanks. :-) I'll save your second suggestion here as well:

trace add variable ::tcltest::numTests(Failed) write {
    apply {args {
        puts {*** EXIT ON ERROR ***}
        exit 1
    }}
}

[sshine]

Closing this issue, since the Tcl track has been bootstrapped without test generators so far.