[libvirt PATCH] qemu: reject readonly attribute for virtiofs

[Peter Krempa]

On Wed, May 13, 2020 at 10:57:33 +0100, Daniel Berrange wrote:
> On Wed, May 13, 2020 at 11:51:50AM +0200, Peter Krempa wrote:
> > On Wed, May 13, 2020 at 10:41:23 +0100, Daniel Berrange wrote:
> > > On Wed, May 13, 2020 at 11:34:05AM +0200, Andrea Bolognani wrote:
> > > > On Wed, 2020-05-13 at 10:58 +0200, Ján Tomko wrote:
> > > > > This is not yet supported by virtiofsd.
> > > > > 
> > > > > Fixes #23
> > > > 
> > > > Please include the full URL here:
> > > > 
> > > >   https://gitlab.com/libvirt/libvirt/-/issues/23
> > 
> > While I definitely don't like gitlab's URIs ...
> > 
> > > Using the '#23' format is the recommended way. GitLab UI will
> > > turn it into a hyperlink, and automatically close the mentioned
> > > issue.
> > 
> > ... just mentioning #23 is useless for users looking at the git repo
> > directly (not through the web-ui) or if we at some point decide to
> > abandon gitlab.
> > 
> > Even if gitlab can't handle the automation of closing the comment when
> > the URI is mentioned I don't see why we should make it harder for users
> > who want to stay away from the browser as much as possible.
> 
> The point is for libvirt to follow normal practice from GitLab, so that
> contributors don't have to know about libvirt specific rules for contributing
> to the project. Telling users to change the normal issue syntax into a URL
> whenever they send a patch is not something we should be doing.

Even if they are flawed?! Just mentioning a random number in the commit
message may be nice for people looking at the code via web-ui. That is
nice but you can't do much there. But I and many other look at the code
in the local checkout and 'git' doesn't randomly decide to expand the
number to the gitlab uri and make it clickable.

Now if I'd like to look at the issue after some time I'll e.g. when
going through git-blame I'll have to open the browser, open gitlab, find
the project and find the issue rather than just click a link. That's
stupid. If gitlab can't deal with the link, it's frankly broken and we
should not give in to a broken approach just because everybody else
does.