[ofa-general] Re: [PATCH ibverbs] Make the gid argument to ibv_attach_mcast and ibv_detach_mcast const

[Jason Gunthorpe]

On Wed, Jul 22, 2009 at 12:02:18PM +0300, Jack Morgenstein wrote:
> On Monday 20 July 2009 21:53, Jason Gunthorpe wrote:
> > I have also patches for mlx4 and mthca to suppress the compiler
> > warning that results from this patch. ipath is OK as is, and I'm not
> > sure where the iwarp stuff lives..
 
> Is this change really necessary?

This makes it easier/clearer for user apps - in C++ constness errors
are not warnings.

> Seems to me that you are creating cross-library dependencies (at
> least with regard to compiler warnings) -- the user will now be
> required to change the device libraries if libibverbs is upgraded.

What? No, not at all. This change is ABI neutral in all the binary
interfaces. All that happens is you get harmless compile warnings in
the device libraries.

The patches for the device libraries are just to remove the harmless
warning, it doesn't affect anything.

> Also, if the device libraries get changed, but libibverbs stays the
> same, I think you will also get warnings if anyone tries to compile
> user-level apps.

Nobody will get warnings from existing user-level apps. The compiler
silently casts from non-const to const.

Jason