Xpra: Ticket #392: integrate a win32 crash report tool

On Linux, we can more easily get gdb backtraces, even from users - see debugging with gdb. But on win32, by the time we get a user report about a crash, all we are likely to hear is "it crashed, dr watson came up".

We ought to be able to do better than that. Maybe using:

For a start, we can collect more data. The log file basic would be a start.



Thu, 17 Oct 2013 07:43:15 GMT - Antoine Martin: owner, status, milestone changed

too late for 0.11


Mon, 20 Jan 2014 11:12:41 GMT - Antoine Martin: milestone changed

I think that the win32 client has been very stable, which means this is a less of a priority.


Sun, 03 Aug 2014 06:17:25 GMT - Antoine Martin: attachment set

shows the alert box for a command line parsing error on the "quality" option


Sun, 03 Aug 2014 06:17:55 GMT - Antoine Martin: attachment set

shows an alert box with some help text for a "help" command line option


Sun, 03 Aug 2014 06:22:17 GMT - Antoine Martin:

As of r7100, we now have a catch-all alert box shown on win32 when the client exits because of an error. (this can be disabled by setting XPRA_MESSAGEBOX=0)

This does not apply to things like network disconnections which still use a non-zero exit code, this is only for bugs or other serious and unexpected problems.

Some examples:

shows an alert box with some help text for a

shows the alert box for a command line parsing error on the


Tue, 19 Aug 2014 06:57:55 GMT - Antoine Martin: milestone changed

I haven't seen any hard crashes (the useless dr watson) for a very long time. So I don't think we need a crash handler.

What we could do however is:


Thu, 04 Dec 2014 17:46:40 GMT - Antoine Martin:

related: #752


Mon, 02 Feb 2015 09:54:22 GMT - Antoine Martin: status changed; resolution set

Now that we have remote logging (see #786), I think this is good enough. (bar #637)


Thu, 30 Apr 2015 15:18:41 GMT - Antoine Martin: milestone changed

(update to correct milestone)


Sat, 23 Jan 2021 04:54:03 GMT - migration script:

this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/392