Re: [patch 1/17] Infrastructure to mark exported symbols as unused-for-removal-soon

[Andrew Morton <akpm@osdl.org>]

Arjan van de Ven <arjan@linux.intel.com> wrote:
>
>  As discussed on lkml before; the patch with the infrastructure to deprecate unused symbols
> 
>  This is patch one in a series of 17; to not overload lkml the other 16 will be mailed direct;
>  people who want to see them all can see them at http://www.fenrus.org/unused

A lot of these patches go through major APIs and seemingly-randomly prepare
to unexport things based on whether they are presently used within modules.

So, for example, drivers/base/attribute_container.c gets a whole pile of
exports scheduled for removal, regardless of whether the resulting module
API makes *sense*.  Ditto scsi core.  And lib/*.

For example this:

  EXPORT_SYMBOL(generic_getxattr);
 -EXPORT_SYMBOL(generic_listxattr);
 +EXPORT_UNUSED_SYMBOL(generic_listxattr); /* removal in 2.6.19 */
  EXPORT_SYMBOL(generic_setxattr);
  EXPORT_SYMBOL(generic_removexattr);

just seems random to me, and it's setting us up for later churn.

So hum.  Don't you think it'd be better to look at each API as a whole,
make decisions about what parts of it _should_ be offered to modules,
rather then looking empirically at which parts presently _need_ to be
exported?