On 25/05/17 14:35 +0200, Richard Biener wrote:
On May 25, 2017 1:38:36 PM GMT+02:00, Nathan Sidwell <nat...@acm.org> wrote:
On 05/25/2017 07:21 AM, Jonathan Wakely wrote:
I don't mind removing the warning again if preferred. I thought it
was
useful (as we already warn for ignored const in return types).
Oh yeah, I recall noticing we did that (and noting we didn't warn
elsewhere). This new warning seems consistent.
I say leave it in unless the grumbling gets too much for you :)
I wonder if we can somehow default to -Wno-error=xyz for such kind of 'style'
warnings... Adding const can't possibly break anything or result in wrong
expectations, can it?
Now that G++ correctly ignores the const it can't change (or break)
anything to add const in the cast.
Before I fixed PR 80544, the presence/absence of the const affected
the generated code and could result in e.g. different overloaded
functions being called (in some fairly obscure cases).
The original report I got was http://ideone.com/JSFEZ3 and GCC was
giving different behaviour to all other C++ compilers (which ignored
the const and so failed the static assertion).
The warning is just saying "hey, you know what you wrote is going to
be ignored, right?" That's a bit like "statement has no effect"
warnings, although those warnings are usually because you mistyped
something and so find real bugs.
