Fabiano Rosas <[email protected]> writes:
> Markus Armbruster <[email protected]> writes:
[...]
>> The entire parser should be burned with fire.
>>
>> What are the remaining differences to the keyval.c parser after your
>> series? Can we deprecate them?
>>
>
> I ran a few of the qemu-opts tests with the keyval code and it differs
> mostly in the handling of empty (non-implied) key and implied key +
> empty value. I'll take a close look next week but I think we should
> deprecate them anyway.
>
> It looks easy to add a flag to keyval_parse_one() to turn its errors
> into deprecation warnings ("warning: madness deprecated"). We could then
> gradually make the conversion without having to wait for the deprecated
> parts to be removed. The child is almost 10 years old.
>
> d454dbe0ee3 ("keyval: New keyval_parse()")
> Author: Markus Armbruster <[email protected]>
> Date: Tue Feb 28 22:26:49 2017 +0100
Worth a try, I think.
Thanks!
[...]
