Discussion: Interpretation of system property flags

Daniel Fuchs daniel.fuchs at oracle.com
Wed Dec 4 17:17:19 UTC 2024


Hi Eirik,

Yes, the situation we're in is far from perfect. I've never been a big
fan of considering that an empty value means "true": if you're going
to pass a System property, you can go the extra mile and supply
its value too, which should make the intent uncontroversial. Note that
not passing the property doesn't necessarily means that the value
is false, since some properties may have a default value which is
"true".

Practically, if we wanted to change such long standing behaviors
in order to align on a new consistent handling for all the boolean
properties we probably would have to "deprecate" at least some
of the current properties and introduce new property names
for them. We would end up with both old names and new names and
therefore would still have to live with the inconsistency
(and possibly increased confusion).

So I agree with Roger on this, the payback might not be worth it.

best regards,

-- daniel

On 04/12/2024 15:03, Roger Riggs wrote:
> Common utility methods might be a small code cleanup but aligning all 
> uses might have a very small payback for the possible incompatibilties.
> 
> Regards, Roger



More information about the core-libs-dev mailing list