Re: [PULL 31/31] qemu-option: warn for short-form boolean options

[Paolo Bonzini]

On 16/02/21 12:04, Peter Maydell wrote:
> On Tue, 16 Feb 2021 at 10:43, Paolo Bonzini <pbonzini@redhat.com> wrote:
> > For this to work, one would need to have a typed string->QAPI parser,
> > i.e. one that takes the schema as input rather than doing a generic
> > parsing to QDict and then using the schema via the visitor.  That would
> > IMHO be overengineered for the purpose of saving five keystrokes on
> > "server,nowait".
>
> The point is not that it's saving five keystrokes. The point
> is that "server,nowait" is how we've documented that users
> should select that functionality since forever, and indeed
> is how we're still documenting it today, and therefore changing
> it is breaking the existing working setups of a very large
> group of users.
>
> Historical warts that nobody much was really using can be safely
> excised; historical warts with many many users much less so.

I agree, and that's why I have no plans to move -chardev off QemuOpts; 
warning is a different step than excising and sometimes years pass from 
one to the other.  However, that doesn't prevent introducing a warning 
so that users slowly move away from the problematic functionality.

Also, "-chardev" is not the way that most users will configure sockets. 
 The more common "-serial tcp:localhost:12345,server,nowait" will not 
trigger the warning; that was not clear at all from the the commit 
message.  It may even make sense to deprecate the *long* form in that 
case (which I am not planning to do, to be clear).   I'll fix the 
documentation for those uses that are affected, though; thanks for 
reporting that.

Paolo