Re: [Emacs-diffs] master 716b468: Extend `file-notify-test02-rm-watch'

[Michael Albinus <michael.albinus@gmx.de>]

Ken Brown <kbrown@cornell.edu> writes:

> Hi Michael,

Hi Ken,

> On my systems I seem to need a much larger sit-for timeout (on the
> order of 0.1 or more rather than 0.001) in order for the test to pass.
> But maybe this is not worth worrying about, since it's clear that this
> is a timing issue rather than a file-notify problem.

I've moved the `sit-for' call to `file-notify--test-read-event' by
intention, so we have just one place to change. You could do something
like

(if (eq system-type 'cygwin)
  (sit-for 0.1 'nodisp))

The other backends do not need them as far as I'm aware of, so do it at
your convenience. Maybe also with a comment ...

>> One point still irritates me: the tests run under cygwin change the
>> results after a while. Several tests which pass after a reboot of a
>> machine, will fail when I use the machine for some hours. It isn't the
>> first time I've observed this, so I need to reboot my machine again and
>> again when working for cygwin.
>>
>> Do you know this behaviour?
>
> No, I don't see this.  I hardly ever reboot my computer.  But I mostly
> run 64-bit Cygwin.  Are you running 32-bit Cygwin?  And are the
> problems related to fork failures?  (You would see error messages to
> this effect on the terminal or in the logs.)  If so, I might be able
> to help.  But in that case we should probably continue the discussion
> off-list.

Likely 32-bit. But I use it very rarely, so I don't want to waste your
time with debugging when you don't know it from somewhere else.

> Regards,
>
> Ken

Best regards, Michael.