Question about ABORT when canceling OPP file transfer

Question about ABORT when canceling OPP file transfer

[ext-jablonski.radoslaw]

Hi Johan,

There is one PTS test for OPP which tests if we're disconnecting cleanly 
when cancelling outgoing file transfer (this test is not mandatory). 
Currently it fails, because we don't send ABORT before disconnect. 

I've noticed that code for handling it nicely already exist in 
gw_obex_xfer_do_abort(...) but is disabled by default (via 
USE_NICE_ABORT definition).

Is it safe to change that and  use "nice abort" by default? Were there problems 
with handling this ABORT command by some devices in described scenario?

BR,
Radek

Re: Question about ABORT when canceling OPP file transfer

[Luiz Augusto von Dentz]

Hi Radek,

On Wed, May 25, 2011 at 10:39 AM,  <ext-jablonski.radoslaw@nokia.com> wrote:
> Hi Johan,
>
> There is one PTS test for OPP which tests if we're disconnecting cleanly
> when cancelling outgoing file transfer (this test is not mandatory).
> Currently it fails, because we don't send ABORT before disconnect.

Just to confirm, this test is for client, right?

> I've noticed that code for handling it nicely already exist in
> gw_obex_xfer_do_abort(...) but is disabled by default (via
> USE_NICE_ABORT definition).
>
> Is it safe to change that and  use "nice abort" by default? Were there problems
> with handling this ABORT command by some devices in described scenario?

Well if you want to have this test to pass obviously you need to send
abort properly, in the other hand I think we should make this a proper
API on gwobex so the application using it can pass to
gw_obex_xfer_do_abort how it wants to abort.

Regards,

-- 
Luiz Augusto von Dentz
Computer Engineer

RE: Question about ABORT when canceling OPP file transfer

[Tom Allebrandi]

Hi All -

> On Wed, May 25, 2011 at 10:39 AM,  <ext-jablonski.radoslaw@nokia.com>
> wrote:
> > Hi Johan,
> >
> > There is one PTS test for OPP which tests if we're disconnecting
> > cleanly when cancelling outgoing file transfer (this test is not
mandatory).
> > Currently it fails, because we don't send ABORT before disconnect.
> 
> Just to confirm, this test is for client, right?
> 
> > I've noticed that code for handling it nicely already exist in
> > gw_obex_xfer_do_abort(...) but is disabled by default (via
> > USE_NICE_ABORT definition).
> >
> > Is it safe to change that and  use "nice abort" by default? Were there
> > problems with handling this ABORT command by some devices in described
> scenario?
> 
> Well if you want to have this test to pass obviously you need to send
abort
> properly, in the other hand I think we should make this a proper API on
> gwobex so the application using it can pass to gw_obex_xfer_do_abort how
> it wants to abort.
> 

I think he is referring to TC_CLIENT_OPH_BV_27_I -- "To verify that a client
can ABORT an OBEX push operation."

This is an optional (category X) test case that was added to exercise the
OBEX ABORT command. It is not required to pass this test case for
qualification.

It's a fairly new test case, it was just added in 4.1 or 4.2 of PTS.

Cheers!

--- tom
tom allebrandi
(Bluetooth SIG PTS Development Team: OPP, FTP, BIP, DUN, HCRP)
wyrles@ytram.com

Re: Question about ABORT when canceling OPP file transfer

[Johan Hedberg]

Hi Radek,

On Wed, May 25, 2011, ext-jablonski.radoslaw@nokia.com wrote:
> There is one PTS test for OPP which tests if we're disconnecting
> cleanly when cancelling outgoing file transfer (this test is not
> mandatory).  Currently it fails, because we don't send ABORT before
> disconnect. 
> 
> I've noticed that code for handling it nicely already exist in
> gw_obex_xfer_do_abort(...) but is disabled by default (via
> USE_NICE_ABORT definition).
> 
> Is it safe to change that and  use "nice abort" by default? Were there
> problems with handling this ABORT command by some devices in described
> scenario?

I think this is just a packaging bug. There's no reason to disable the
"nice abort" feature. Actually, I think it'd be fine not to have this
configurable for gwobex but always enable it (it's a legacy from really
old OpenOBEX versions that didn't yet support sending the Abort
command).

Johan