[jitsi-dev] strange (protocol) messages when writing to vtok client on iPhone


#1

Hi

After not finding anything about this (or a similar) issue on the mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an iPhone there are some "protocol" messages before and after each written text-message on the iphone looking like "A5BF" or similar. Seems like an increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

     STefan

PS: using latest versions of jitsi and vtok.


#2

Sounds like this could be coming from OTR. Not sure why though.

Emil

···

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

     STefan

PS: using latest versions of jitsi and vtok.

--
http://jitsi.org


#3

Anything I can do to help debugging analyze or correct the issue?

Regards

     STefan

···

On 20.03.12 12:55, Emil Ivov wrote:

Sounds like this could be coming from OTR. Not sure why though.

Emil

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

      STefan

PS: using latest versions of jitsi and vtok.


#4

Is this occurring with a specific peer? Can you check if in the OTR menu
you have Automatic Chat Encryption checked for this peer?

Emil

···

On 20.03.12 14:07, STefan Mayer wrote:

Anything I can do to help debugging analyze or correct the issue?

Regards

     STefan

On 20.03.12 12:55, Emil Ivov wrote:

Sounds like this could be coming from OTR. Not sure why though.

Emil

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

      STefan

PS: using latest versions of jitsi and vtok.

--
Emil Ivov, Ph.D. 67000 Strasbourg,
Project Lead France
Jitsi
emcho@jitsi.org PHONE: +33.1.77.62.43.30
http://jitsi.org FAX: +33.1.77.62.47.31


#5

Hi

I have only one peer (my wife :wink: where I can test.... but encryption is off (manually) for that peer... I also turned off private-messaging now for that peer, but still the same issue...

Regards

     STefan

···

On 20.03.12 14:51, Emil Ivov wrote:

Is this occurring with a specific peer? Can you check if in the OTR menu
you have Automatic Chat Encryption checked for this peer?

Emil

On 20.03.12 14:07, STefan Mayer wrote:

Anything I can do to help debugging analyze or correct the issue?

Regards

      STefan

On 20.03.12 12:55, Emil Ivov wrote:

Sounds like this could be coming from OTR. Not sure why though.

Emil

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

       STefan

PS: using latest versions of jitsi and vtok.


#6

Hi Stefan, devs,

I can reproduce this problem on my side: sending messages from Jitsi to Vtalk.

The following scenario corresponds to the capture joined:

1) Vtalk sends the message "Rh" to Jitsi (packet 1).
--> Jitsi displays correctly this message.

2) Jitsi sends the message "jkhkj" to Vtalk (packets 2, 3 and 4). Solely packet 3 contains a <body> with the message "jkhkj", but Vtalk displays 3 messages (cf. screenshot joined):
--> "V14FY1", corresponds to packet 2 <thread> tag content.
--> "jkhkj", corresponds to packet 3 <body> tag content.
--> "V14FY3", corresponds to packet 4 <thread> tag content.

In fact, packets 2 and 4 are used to send typing notification.

Thus (just an hypothesis, but corresponds to our situation), when Vtalk receives a message, it seems to display the first tag contained in the <message> stanza. Does anyone knows a XEP that defines that the <body> element must always be present and placed as the first tag contained by a <message> stanza?

- If yes, then we have to add an empty <body/> tag in our typing notification messages.

- If no, we may inform Vtalk developers.

Regards,
Vincent

P.S.: I do not know how the message <thread> work in practice, but is it normal to have the thread ID incremented for each new <message> stanza?

vtalk_thread_problem.pcap (1.22 KB)

···

On 03/21/2012 08:40 AM, STefan Mayer wrote:

Hi

I have only one peer (my wife :wink: where I can test.... but encryption is
off (manually) for that peer... I also turned off private-messaging now
for that peer, but still the same issue...

Regards

STefan

On 20.03.12 14:51, Emil Ivov wrote:

Is this occurring with a specific peer? Can you check if in the OTR menu
you have Automatic Chat Encryption checked for this peer?

Emil

On 20.03.12 14:07, STefan Mayer wrote:

Anything I can do to help debugging analyze or correct the issue?

Regards

STefan

On 20.03.12 12:55, Emil Ivov wrote:

Sounds like this could be coming from OTR. Not sure why though.

Emil

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an
iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems
like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

STefan

PS: using latest versions of jitsi and vtok.

--
Vincent Lucas, Ph.D. Jitsi developer
chenzo@jitsi.org http://jitsi.org


#7

Hi

Yes this seems to be exactly the issue I encounter, except I couldn't capture it yet...

Also I have absolutely no clue how to solve it... hopefully the devs will find a solution...

Thanks a lot for your help, and a great piece of software!

Regards

     STefan

···

On 23.03.12 11:08, Vincent Lucas wrote:

Hi Stefan, devs,

I can reproduce this problem on my side: sending messages from Jitsi to Vtalk.

The following scenario corresponds to the capture joined:

1) Vtalk sends the message "Rh" to Jitsi (packet 1).
--> Jitsi displays correctly this message.

2) Jitsi sends the message "jkhkj" to Vtalk (packets 2, 3 and 4). Solely packet 3 contains a <body> with the message "jkhkj", but Vtalk displays 3 messages (cf. screenshot joined):
--> "V14FY1", corresponds to packet 2 <thread> tag content.
--> "jkhkj", corresponds to packet 3 <body> tag content.
--> "V14FY3", corresponds to packet 4 <thread> tag content.

In fact, packets 2 and 4 are used to send typing notification.

Thus (just an hypothesis, but corresponds to our situation), when Vtalk receives a message, it seems to display the first tag contained in the <message> stanza. Does anyone knows a XEP that defines that the <body> element must always be present and placed as the first tag contained by a <message> stanza?

- If yes, then we have to add an empty <body/> tag in our typing notification messages.

- If no, we may inform Vtalk developers.

Regards,
Vincent

P.S.: I do not know how the message <thread> work in practice, but is it normal to have the thread ID incremented for each new <message> stanza?

On 03/21/2012 08:40 AM, STefan Mayer wrote:

Hi

I have only one peer (my wife :wink: where I can test.... but encryption is
off (manually) for that peer... I also turned off private-messaging now
for that peer, but still the same issue...

Regards

STefan

On 20.03.12 14:51, Emil Ivov wrote:

Is this occurring with a specific peer? Can you check if in the OTR menu
you have Automatic Chat Encryption checked for this peer?

Emil

On 20.03.12 14:07, STefan Mayer wrote:

Anything I can do to help debugging analyze or correct the issue?

Regards

STefan

On 20.03.12 12:55, Emil Ivov wrote:

Sounds like this could be coming from OTR. Not sure why though.

Emil

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an
iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems
like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

STefan

PS: using latest versions of jitsi and vtok.


#8

Hi

Any progress on this issue? It still seems to be occuring in the latest dev-versions...

Can I help somehow?

regards

     STefan

···

On 23.03.12 14:26, STefan Mayer wrote:

Hi

Yes this seems to be exactly the issue I encounter, except I couldn't capture it yet...

Also I have absolutely no clue how to solve it... hopefully the devs will find a solution...

Thanks a lot for your help, and a great piece of software!

Regards

    STefan

On 23.03.12 11:08, Vincent Lucas wrote:

Hi Stefan, devs,

I can reproduce this problem on my side: sending messages from Jitsi to Vtalk.

The following scenario corresponds to the capture joined:

1) Vtalk sends the message "Rh" to Jitsi (packet 1).
--> Jitsi displays correctly this message.

2) Jitsi sends the message "jkhkj" to Vtalk (packets 2, 3 and 4). Solely packet 3 contains a <body> with the message "jkhkj", but Vtalk displays 3 messages (cf. screenshot joined):
--> "V14FY1", corresponds to packet 2 <thread> tag content.
--> "jkhkj", corresponds to packet 3 <body> tag content.
--> "V14FY3", corresponds to packet 4 <thread> tag content.

In fact, packets 2 and 4 are used to send typing notification.

Thus (just an hypothesis, but corresponds to our situation), when Vtalk receives a message, it seems to display the first tag contained in the <message> stanza. Does anyone knows a XEP that defines that the <body> element must always be present and placed as the first tag contained by a <message> stanza?

- If yes, then we have to add an empty <body/> tag in our typing notification messages.

- If no, we may inform Vtalk developers.

Regards,
Vincent

P.S.: I do not know how the message <thread> work in practice, but is it normal to have the thread ID incremented for each new <message> stanza?

On 03/21/2012 08:40 AM, STefan Mayer wrote:

Hi

I have only one peer (my wife :wink: where I can test.... but encryption is
off (manually) for that peer... I also turned off private-messaging now
for that peer, but still the same issue...

Regards

STefan

On 20.03.12 14:51, Emil Ivov wrote:

Is this occurring with a specific peer? Can you check if in the OTR menu
you have Automatic Chat Encryption checked for this peer?

Emil

On 20.03.12 14:07, STefan Mayer wrote:

Anything I can do to help debugging analyze or correct the issue?

Regards

STefan

On 20.03.12 12:55, Emil Ivov wrote:

Sounds like this could be coming from OTR. Not sure why though.

Emil

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an
iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems
like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

STefan

PS: using latest versions of jitsi and vtok.


#9

Not so far.

Still on our todo list though!

Could you please open a ticket on our tracker so that we don't forget
about it?

Cheers,
Emil

···

On 12.04.12 13:44, STefan Mayer wrote:

Hi

Any progress on this issue? It still seems to be occuring in the latest
dev-versions...

Can I help somehow?

regards

     STefan

On 23.03.12 14:26, STefan Mayer wrote:

Hi

Yes this seems to be exactly the issue I encounter, except I couldn't
capture it yet...

Also I have absolutely no clue how to solve it... hopefully the devs
will find a solution...

Thanks a lot for your help, and a great piece of software!

Regards

    STefan

On 23.03.12 11:08, Vincent Lucas wrote:

Hi Stefan, devs,

I can reproduce this problem on my side: sending messages from Jitsi
to Vtalk.

The following scenario corresponds to the capture joined:

1) Vtalk sends the message "Rh" to Jitsi (packet 1).
--> Jitsi displays correctly this message.

2) Jitsi sends the message "jkhkj" to Vtalk (packets 2, 3 and 4).
Solely packet 3 contains a <body> with the message "jkhkj", but Vtalk
displays 3 messages (cf. screenshot joined):
--> "V14FY1", corresponds to packet 2 <thread> tag content.
--> "jkhkj", corresponds to packet 3 <body> tag content.
--> "V14FY3", corresponds to packet 4 <thread> tag content.

In fact, packets 2 and 4 are used to send typing notification.

Thus (just an hypothesis, but corresponds to our situation), when
Vtalk receives a message, it seems to display the first tag contained
in the <message> stanza. Does anyone knows a XEP that defines that
the <body> element must always be present and placed as the first tag
contained by a <message> stanza?

- If yes, then we have to add an empty <body/> tag in our typing
notification messages.

- If no, we may inform Vtalk developers.

Regards,
Vincent

P.S.: I do not know how the message <thread> work in practice, but is
it normal to have the thread ID incremented for each new <message>
stanza?

On 03/21/2012 08:40 AM, STefan Mayer wrote:

Hi

I have only one peer (my wife :wink: where I can test.... but encryption is
off (manually) for that peer... I also turned off private-messaging now
for that peer, but still the same issue...

Regards

STefan

On 20.03.12 14:51, Emil Ivov wrote:

Is this occurring with a specific peer? Can you check if in the OTR
menu
you have Automatic Chat Encryption checked for this peer?

Emil

On 20.03.12 14:07, STefan Mayer wrote:

Anything I can do to help debugging analyze or correct the issue?

Regards

STefan

On 20.03.12 12:55, Emil Ivov wrote:

Sounds like this could be coming from OTR. Not sure why though.

Emil

On 19.03.12 10:39, STefan Mayer wrote:

Hi

After not finding anything about this (or a similar) issue on the
mailing lists and other sources I'll explain it here:

When using Jitsi on my MacBook pro writing to a vtok client on an
iPhone
there are some "protocol" messages before and after each written
text-message on the iphone looking like "A5BF" or similar. Seems
like an
increasing "message identifier" or similar.

It only seems to be the combination Jitsi <=> vtok on iPhone , as I
tried various other combinations, never having this problem...

Anyone else seen this?

Kind regards

STefan

PS: using latest versions of jitsi and vtok.

--
Emil Ivov, Ph.D. 67000 Strasbourg,
Project Lead France
Jitsi
emcho@jitsi.org PHONE: +33.1.77.62.43.30
http://jitsi.org FAX: +33.1.77.62.47.31