Support Unimplemented attribute #826
Conversation
swernli
requested review from
billti,
cesarzc,
minestarks,
idavis,
DmitryVasilevsky and
ScottCarda-MS
as code owners
|
whoops... looks like I somehow merged @DmitryVasilevsky 's changes with my own here. I'll try to fix this up... |
swernli
force-pushed
the
swernli/unimplemented-attr
branch
from
d3d32fe to
97a5024
Compare
There was a problem hiding this comment.
This implementation LGTM. Thank you!
Since we had that discussion about the fate of @Config() recently, looking at this PR with fresh eyes, this feature seems more-or-less equivalent to the @Config() attribute. We could define a special Config expression that should never be true, like @Config(Unimplemented). Would reduce complexity (since it would just get handled along with all the other @Configged items in the preprocessing step) and I think the behavior would be ... the same? Minus some very specific error reporting... which could be worth the added complexity.
That's my current take on the feature itself, but I'm happy whichever way you decide to go. Thanks for all the revisions and discussion!
This change introduces the @unimplemented() attributed which can be used to mark an item as something that should not be used. The motivation behind including this and not just treating it as a Not Found error is to give the user more information that this API may have previously existing and is not currently supported (see #699 for context).
Co-authored-by: Scott Carda <[email protected]>
swernli
force-pushed
the
swernli/unimplemented-attr
branch
from
c823d7c to
d2fa415
Compare
This change introduces the
@Unimplemented()attributed which can be used to mark an item as something that should not be used. The motivation behind including this and not just treating it as aNot Founderror is to give the user more information that this API may have previously existing and is not currently supported (see #699 for context).