Richi Plana wrote:
> On Sun, 2007-12-30 at 21:48 +0000, Colin Guthrie wrote:
>> Richi Plana wrote:
>>> It doesn't tell ME much, save that the "Stream died" (for who-knows-what
>>> reason). One thing I did notice that was different was that on viper, PS
>>> kept spouting out the line "Checking for dead streams ..." that duke did
>>> not.
>> Yup "stream died" at the client end, but it corresponds to the "protocol
>> error" at the server end.
>>
>> So something isn't working with regards to my patches :(
>>
>> It worked OK for me, but that's not overly helpful for you.
>>
>> Unless Lennart pops up to correct #192 (the ticket to which I attached
>> my patch - unless it was #193 - I can't remember right now :)) then I'll
>> try to do more testing myself in the next few days (tho' Jan 1st is a
>> right off in terms of concious thought :))
>
> Thanks. I just wish I could be more involved. This is what I found out,
> though. On the "server" side,
> pulsecore/protocol-native.c:protocol_error() gets called first from this
> part:
>
> if (!pa_tagstruct_eof(t)) {
> protocol_error(c);
> return;
> }
>
> around line 1418 of protocol-native.c. It then gets called 3 times from:
>
> if (!pa_tagstruct_eof(t)) {
> protocol_error(c);
> return;
> }
>
> around line 1246 of protocol-native.c.
>
> Anything I can do to help debug?
Well I've just been poking around myself and finding my way. There is a
bit in the code where module-tunnel.c where for protocol versions >=12
add some extra flags....
Actually can you check to see if any of your other patches changes
module-tunnel.c - one of the older patches hacked the value of
PROTOCOL_VERSION to the hardcoded value of 11 - if so please don't apply
this patch as my one supercedes it (although I don't think this would
actually break things, I could be wrong)
Col
