Re: [PATCH v2 2/5] libxl: vcpuset: Return error values.

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Wed, Mar 18, 2015 at 02:09:09PM +0000, Dario Faggioli wrote:
> On Wed, 2015-03-18 at 13:08 +0000, Ian Campbell wrote:
> > On Wed, 2015-03-18 at 13:06 +0000, Ian Campbell wrote:
> > > On Fri, 2015-03-13 at 16:26 -0400, Konrad Rzeszutek Wilk wrote:
> > > > The function does not return any values at all. Convert the
> > > > internal libxl ones (ERROR_FAIL, ..., etc) to positive values
> > > > and for the other cases just return standard libxl values.
> > > 
> > > It's not clear why you want to do this, in particular returning
> > > -ERROR_INVAL and inverting libxl error codes seems like a very strange
> > > thing to be doing.
> > 
> > BTW I know the xl error handling is horribly confused, and there are
> > even a small number of instances of -ERROR_* already, but I think those
> > are wrong and we shouldn't introduce more.
> > 
> Indeed. I did some xl error code refactoring for a series of mine a few
> days back, and as far as I could see, the most common pattern in xl is
> returning 0 or 1.

Gah, I seem to have looked at the wrong examples and thought that was
the proper way!
> 
> FWIW, I think we should not diverge any further from that and, at some
> point, convert 0/1 to EXIT_SUCCESS/EXIT_FAILURE.
> 
> > > I think you should either use ERROR_INVAL (not inverted) and propagate
> > > libxl rc's directly or convert them into something which suits xl, i.e.
> > > 0 and 1.
> > > 
> Again, +1 for 0 or 1.
> 
> Regards,
> Dario