On Thu, Jan 29, 2015 at 1:18 AM, Kevin DuBois
<email address hidden> wrote:
> Review: Needs Fixing
>
> int mir_connection_get_fd(MirConnection* connection);
> It seems more important what the fd is for, so I'd rather the fn say
> something about what the fd does... maybe:
> int mir_connection_event_dispatch_fd(MirConnection* connection)
Seems reasonable, I'll use that naming.
>
>
> 201 + thread_local static bool running;
> Nothing wrong here, but we still haven't properly sorted out TLS with
> hybris. We'd probably be a bit more robust if we still avoid using
> TLS.
On Thu, Jan 29, 2015 at 1:18 AM, Kevin DuBois get_fd( MirConnection* connection); event_dispatch_ fd(MirConnectio n* connection)
<email address hidden> wrote:
> Review: Needs Fixing
>
> int mir_connection_
> It seems more important what the fd is for, so I'd rather the fn say
> something about what the fd does... maybe:
> int mir_connection_
Seems reasonable, I'll use that naming.
>
>
> 201 + thread_local static bool running;
> Nothing wrong here, but we still haven't properly sorted out TLS with
> hybris. We'd probably be a bit more robust if we still avoid using
> TLS.
Boo. And it was sooooo elegant.
I'll rework this to not need TLS.
>
> 704 + if (pipe2(pipefds, O_NONBLOCK) < 0)
> O_CLOEXEC?
While we're at it, yes.