[RFC/Patch] libstdc++/14320 aka class type streamoff doesn't work :(

[Pétur Runólfsson]

Paolo Carlini wrote:
> Thus the below: it seems to me that the best we can do for streamoff is
> essentially returning to what we had in 3_3. This implies giving away
> some type safety (i.e., some QoI) and also that the issue with LFS is
> still there, but otherwise seems OK. On glibc platforms we get an XPASS,
> since int64t is actually a typedef for long long.

This patch seems reasonable to me given the current release plan for
3.4.

A minor nitpick:

@@ -204,7 +111,7 @@
       // conversion. In this implementation this constructor stores
       // the integer as the offset and default constructs the state.
       /// Construct position from integer.
-      fpos(__streamoff_base_type __off)
+      fpos(int __off)
       : _M_off(__off), _M_state() { }
 
       // The standard requires that fpos objects can be constructed
@@ -214,9 +121,12 @@
       // constructor stores the streamoff as the offset and default
       // constructs the state.
       /// Construct position from offset.
-      fpos(const streamoff& __off)
+      fpos(streamoff __off)
       : _M_off(__off), _M_state() { }

If streamoff is an integral type, it should be sufficient to have only
the second constructor.

> What do you think? Disabling at build time LFS, I don't believe this is
> an option,

I agree.

Regards,
Petur