[Issue 10126] New: Make TaskPool terminate on its own or improve docs to make it clear that it won't

d-bugmail at puremagic.com d-bugmail at puremagic.com
Tue May 21 01:00:09 PDT 2013


http://d.puremagic.com/issues/show_bug.cgi?id=10126

           Summary: Make TaskPool terminate on its own or improve docs to
                    make it clear that it won't
           Product: D
           Version: D2
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Phobos
        AssignedTo: nobody at puremagic.com
        ReportedBy: jmdavisProg at gmx.com


--- Comment #0 from Jonathan M Davis <jmdavisProg at gmx.com> 2013-05-21 01:00:03 PDT ---
This program will never exit:

import std.parallelism;
import std.stdio;

void main()
{
    auto pool = new TaskPool;
    writeln("Exiting...");
}

It prinst "Exiting..." and then just sits there. After poking at this for a
couple of hours, I finally figured out that you have to call finish on the
TaskPool to get it to terminate. My first inclination is that the threads
should properly terminate when the main thread terminates and that it's a bug
that they don't, but it may very well be that they have to work that way for
one reason or another. But the docs are _not_ clear on this point. After
digging through the docs in std.parallelism, I see some mention in stuff like
taskPool of needing to call finish normally, but the main TaskPool docs say
nothing of the sort.

So, please either make it so that the TaskPool's threads terminate when main
does or clarify TaskPool's documentation so that the correct usage is clear.
There probably needs to be an example on TaskPool on its proper usage so that
it's clear that finish needs to be called (as well as any other quirks of which
I'm not aware). Otherwise, it's too easy to miss them and end up with a program
that won't terminate for no obvious reason.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------


More information about the Digitalmars-d-bugs mailing list