On Thu, Jun 02, 2016 at 08:23:37PM +0200, Igor Mammedov wrote: > On Thu, 2 Jun 2016 14:34:27 -0300 > Eduardo Habkost <ehabk...@redhat.com> wrote: [...] > > > > > > > > > > I don't plan to replace plus/minus_features with anything nor to > > > > > make this variables a global ones to spread +- x86/sparc legacy > > > > > format everywhere. > > > > > > > > Can't the +/- semantics be emulated by simply registering > > > > plus_features/minus_features after the other global properties > > > > are registered inside x86_cpu_parse_featurestr()? > > > it could be done, at the first glance it will take 2 extra parsing > > > passes > > > > > > 1: copy featurestr, parse feat=x,feat > > > 2: copy featurestr, parse +feat > > > 3: copy featurestr, parse -feat > > > > Why? Can't we just replace plus_features and minus_features with > > two string lists (or a QDict), and make the corresponding > > object_property_parse()/qdev_prop_register_global() calls after > > the main parsing loop? > > > > (Didn't you do that in your old "target-i386: set [+-]feature > > using static properties" patch?) > ok, I'll do that.
Thanks! [...] > > But as I said above: if we are not deleting any code (and are > > adding extra code instead), I don't see the point of forcibly > > disabling it. We can just leave it there and print a warning. > ok, lets go with warning, saying" > > "[+-]feature syntax is obsoleted and will be removed in future, > it's recommended to use canonical property syntax 'feature=value'" We could print "+%s is obsolete [...] use '%s=on'" and "-%s is obsolete [...] use '%s=off'", depending on the case. This way, users can see more easily what's the proper syntax. -- Eduardo
