Taint-check can actually determine this on its own, but this way the
error messages are more concise because they don't go into the details
of why the return value is tainted.
Note that the return value may actually be safe depending on its type,
but that's not easy for taint-check to know, and a core principle of
taint-check is to prefer false positives to false negatives.
Change-Id: I1e690f535a144cc53cbf3483a8f1d7d8d8eb4519
This makes the limit=… parameter behave more predictable, in my
opinion:
- …?limit=0 is still respected. I don't get why anybody would
do this intentionally. But even if it's an error, ignoring
the number would hide that error.
- …?limit= without a value as well as …?limit fall back to the
default value instead of 0.
Bug: T289351
Change-Id: I9ffa642d6aef235f56253fbf102a4e3943847d31
