[PATCH] Downgrade "definition of dllimport static field" error to warning for class templates (PR19902)

[Nico Rieck]

On 04.06.2014 01:18, Hans Wennborg wrote:
>> If it's perfectly OK and semantically consistent to accept this then why
>> emit a warning?
> 
> For compatibility, I guess. MSVC rejects the code if S is
> instantiated. MinGW errors, but allows it with a warning when using
> -fpermissive. A warning which indicates that the code is in a dodgy
> corner of dllimport land seems valuable.

But curiously they do accept it if S<int>::x is explicitly instantiated.
It's unfortunate that they yet again botched their headers. Not sure if
you were referring to <locale> earlier, but the same issue exists there
(see PR13450).

With which MinGW version did you test? Mine (4.9.0) just drops the
dllimport attribute but this doesn't mean much as they don't seem to
implement any inline semantics for dll attributes.

>> Also note the patch was described as a system header workaround, while
>> you're talking about accepting it as fully valid user code. The correct way
>> to implement this patch depends on which of the two is preferable -- so that
>> needs to be decided on first.
> 
> Yes, I think Reid is suggesting accepting it as valid user code and
> I'm fine with that.

Being more permissive here but making the diagnostic configurable to get
the strict behavior is a nice trade-off IMO.


On 04.06.2014 01:36, Alp Toker wrote:>
> Name this IsValid. The fact it causes a warning is tangential.

Then I'd name it for what it actually represents: whether the variable
is a member of a class or a class template.

-Nico