How do you feel about this API?

/* Channel.Interface.Password */
#define TP_CHANNEL_FEATURE_PASSWORD \
  tp_channel_get_feature_quark_password ()
GQuark tp_channel_get_feature_quark_password (void) G_GNUC_CONST;

void tp_channel_password_provide_password_async (TpChannel *self,
    const gchar *password,
    GAsyncReadyCallback callback,
    gpointer user_data);

gboolean tp_channel_password_provide_password_finish (TpChannel *self,
    GAsyncResult *result,
    GError **error);

gboolean tp_channel_password_needed (TpChannel *self);

/**
 * TpChannel:password-needed:
 *
 * If %TRUE, tp_channel_password_provide_password_async() has to be called
 * to be able to join the channel.
 *
 * This is not guaranteed to be meaningful until tp_proxy_prepare_async() has
 * finished preparing %TP_CHANNEL_FEATURE_PASSWORD; until then, it may return
 * %FALSE even if the channel is actually protected by a password.
 */


Maybe the provide async call should fail with TP_ERROR_AUTHENTICATION_FAILED if the password is not correct?

(In reply to comment #1)
> gboolean tp_channel_password_provide_password_finish (TpChannel *self,
>     GAsyncResult *result,
>     GError **error);


I meant:


gboolean
tp_channel_password_provide_password_finish (TpChannel *self,
    GAsyncResult *result,
    gboolean *correct,
    GError **error)

except, of course, if we decide to fail if the pass was wrong.

(In reply to comment #1)
> /**
>  * TpChannel:password-needed:
>  *
>  * If %TRUE, tp_channel_password_provide_password_async() has to be called
>  * to be able to join the channel.
>  *
>  * This is not guaranteed to be meaningful until tp_proxy_prepare_async() has
>  * finished preparing %TP_CHANNEL_FEATURE_PASSWORD; until then, it may return
>  * %FALSE even if the channel is actually protected by a password.
>  */

For completeness, presumably this emits notify::password-needed when appropriate? (If so, it's worth mentioning that in the docstring.)

> Maybe the provide async call should fail with TP_ERROR_AUTHENTICATION_FAILED if
> the password is not correct?

I like this better than the "out" boolean. I'm not sure why we didn't do this in the D-Bus API...

(Looking at the D-Bus API again, I'm also not sure why ProvidePassword claims to be able to raise InvalidArgument.)

Created attachment 47137 [details] [review]
textchan-group: implement Password

Created attachment 47138 [details] [review]
TpChannel: add high-level API for Password (#37360)

Created attachment 47139 [details] [review]
automatic-proxy-factory: prepare TP_CHANNEL_FEATURE_PASSWORD

Looks good to me.

Merged to master; will be in 0.15.2