[libvirt PATCH] qemu: reject readonly attribute for virtiofs

[Andrea Bolognani]

On Wed, 2020-05-13 at 13:06 +0200, Peter Krempa wrote:
> On Wed, May 13, 2020 at 11:20:07 +0100, Daniel Berrange wrote:
> > It isn't about giving in. Again the point is to not needlessly create
> > special rules for contributing to libvirt, because every special rule
> > we add is another thing for contributors to stumble over. Some rules
> > are worth it because they have meaningful benefits such as the use of
> > Signed-off-by/DCO. The mentioning of full URLs instead of the normal
> > issue reference syntax does not have a meaningful benefit that 
> > justifies a libvirt special rule for contributions.
> 
> I gave an examples of two specific meaningful benefit above:
> 
> 1) it provides a clickable link without second guessing where to go for
> command line users
> 2) provides stable reference to the hosting of issues
> 
> Note that for example github uses exactly the same format for
> referencing issues. That means that it's unclear what we are referring.
> 
[...]
> 
> The shortened issue names are ambiguous and the hosting has no way in
> figuring out where to point to. Providing full URL is not something
> which should be described as "no meaningful benefit" but it actively
> disambiguates the links regardless of where it's hosted or refered from.

I completely agree, #nnn is too ambiguous to be useful.

It seems that GitLab happily accepts full URLs instead of shortened
identifiers:

  https://docs.gitlab.com/ee/user/project/issues/managing_issues.html#default-closing-pattern

We can have a simple prebuild check, similar to the one we already
use for DCO checking, which catches uses of

  Fixes #nnn

and similar and tells contributors to use a full URL instead.

-- 
Andrea Bolognani / Red Hat / Virtualization