[v3] libstdc++/17922

[Paolo Carlini]

Benjamin Kosnik wrote:

>>... and sorry for not not having replied sooner to your hint in the 
>>audit trail: I was sad because of the _S_ios_openmode_end & co bits, 
>>that probably we can't really fix within the current ABI: for v7 maybe 
>>we should bite the bullet and renounce to the enums?!?
>>    
>>
>I think this patch is safe for mainline and gcc-3_4-branch. Adding
>values is not going to cause any problems.
>  
>
Indeed.

>Is there something I just don't see at the moment?
>  
>
No, no, you see clearly everything (as the last entry in the audit trail 
shows ;): I was referring to the problem we have with 
_S_ios_openmode_end & co, which the user does *not* know and should 
*not* trigger a warning if missing from the switch... This is a warning 
regression basically, but unfortunately (to "answer" your question is 
the audit trail) I do *not* have any bright idea to fix it within the 
current ABI... :-(

Paolo.