Re: [msi/tests 3/3] Create a system restore point when needed

[Paul Vriens]

James Hawkins wrote:
> On Sat, Aug 30, 2008 at 3:14 PM, Paul Vriens  wrote:
> > Hi,
> >
> > The slowness on some machines for the install tests comes from the system
> > restore facility. When 'System Restore' is active you will find a great
> > number
> > of system restore points after these tests have run. Adding our own system
> > restore point circumvents this as Windows will now wait till we tell it that
> > the
> >  system change has ended. Only 1 system restore point will be created now
> > and
> > this one is nicely removed afterwards.
> >
> > On my testboxes (XP and Vista) this makes sure that the tests run within the
> > limits of the timeout (2 minutes). Before this patch the tests would take
> > about
> > 8 minutes on my Vista box (now 1.something).
> >
> > Changelog
> >  Create a system restore point when needed
>
> Please factor all of this out into functions.

I'll give that a shot.
> - /* Create a system restore point. We need to do this on a system
>
> where 'System
> -
> * Restore' is active as we would otherwise timeout on these tests.
> -
> * By doing so we circumvent the multitude of restore points that
>
> would have been
> -
> * created by all the installation and removal tests.
> -
> *
> -
> * The restore point is removed afterwards.
> -
> *
> -
> * On systems where 'System Restore' is not active, this won't
>
> have any effect
> -
> * as the first call to SRSetRestorePointA will fail with a status of
> -
> * ERROR_SERVICE_DISABLED.
> -
> */
>
> Does the comment really need to be that long?

I think I can reduce that to a 2-liner or so. During testing I added stuff and 
stuff that's why it's maybe too much.