[PATCH xserver 1/2] xwayland: Move sprite invalidation logic into mipointer

Emil Velikov emil.l.velikov at gmail.com
Mon May 30 20:50:32 UTC 2016


On 29 May 2016 at 19:45, Keith Packard <keithp at keithp.com> wrote:
> Emil Velikov <emil.l.velikov at gmail.com> writes:
>
>>  - Seems like the exports from mi/ are remnants from the separate
>> (non-builtin) extmod days.
>>
>> Not 100% sure though. Can anyone please shed some light ?
>
> _X_EXPORT is intended to mark the boundary between 'internal' bits of
> the server and 'public' bits. So even when not using modules which
> require actual symbol table stuff, it's useful to describe what
> constitutes this interface.
>
> Things marked _X_EXPORT are likely to be used by code which you are not
> running, so changes to them need to be reviewed by people with an eye
> towards preserving compatibility or (at least) catching errors at
> compile time. Interfaces not marked _X_EXPORT are only used within the
> "core server", and so can be changed without as much regard to potential
> incompatibility issues.
>
> In this particular case, the function in question is used by a piece of
> code run in a very specialized environment (Xwayland), so I wanted to
> note that it was part of this special class of functions.
>
So you're saying that even though they are no users of the said
interface it's still nice to define (export) it ? Sure that sounds
perfectly reasonable.

I'm not sure it makes sense to have them if there cannot* be any users
of the said interfaces. You know all the "the binary size will be
smaller", "there are less vectors of attack" kind of arguments.

Thanks
Emil

* I'm not 100% sure if that's the case for !Xorg ddx.


More information about the xorg-devel mailing list