On Fri, Dec 4, 2009 at 10:01 AM, Octavio Alvarez
<
alva...@alvarezp.ods.org> wrote:
> ---
While the title of the patch makes sense in the rebase context, when I
merge this changes nobody will know anything about that old branch, so
they must look as any other change.
So how about this instead:
---
directconn: generic bug fixes
It wasn't actually working before.
---
> cvr/pn_direct_conn.c | 4 ++++
> 1 files changed, 4 insertions(+), 0 deletions(-)
>
> diff --git a/cvr/pn_direct_conn.c b/cvr/pn_direct_conn.c
> index 11e4b48..ae8f53f 100644
> --- a/cvr/pn_direct_conn.c
> +++ b/cvr/pn_direct_conn.c
> @@ -196,6 +196,9 @@ pn_direct_conn_write(struct pn_direct_conn *direct_conn,
> {
> /* Let's write the data. */
> status = pn_stream_write (direct_conn->stream, data, len, &tmp, NULL);
> +
> + if (status == G_IO_STATUS_NORMAL)
> + pn_stream_flush (direct_conn->stream, NULL);
This is not needed unless the channel encoding is changed, which
happens on patch #3. So it would make more sense to join together
patch #2 and #3 to make it a logically atomic change.
> }
>
> if (status == G_IO_STATUS_NORMAL)
> @@ -302,6 +305,7 @@ read_cb(GIOChannel *source, GIOCondition condition, gpointer data)
> {
> pn_error ("failed to allocate memory for read");
>
> + pn_direct_conn_destroy(direct_conn);
> return FALSE;
> }
>
> --
Thanks.
--
Felipe Contreras