Hi,
On 07/05/2013 05:36 PM, Gabriel Dos Reis wrote:
Paolo Carlini <paolo.carl...@oracle.com> writes:
| Hi,
|
| this issue dates back to 2004 and got stalled pretty soon. Essentially
| Gaby wanted to see *dynamic_cast* explicitly mentioned in this kind of
| diagnostic (we used to explicitly talk about static_cast), whereas
| Mark found the current status an improvement over the past, proposed
| some further, minor, improvements (which I tried to implement in the
| below)
|
| Today I noticed that current ICC, clang, OracleStudio, all produce
| diagnostic quite similar, modulo the latter improvements, to GCC and
| decided to explore whether Gaby (and Jason) would like to reassess the
| bug, whether today we want to talk about dynamic_cast.
|
| Tested x86_64-linux.
|
| Thanks,
| Paolo.
I still consider talk of 'static_cast' in this context a non-starter.
However, maybe just saying "convert" would be OK -- though clearly
giving hints of dynamic_cast is much better.
To be clear, currently we do *not* talk of static_cast, likewise the
other compilers I have access to. We say "cannot convert ...".
Minor nits: say "base class" instead of "base", and "derived class"
instead of "derived type".
I see, I wondered whether that would be better.
Otherwise, patch OK with these suggestions.
Great, thanks a lot.
Paolo.
