spectrum crashes in a twitter transport [@ mb_conn_fetch_url_cb]

[GoogleCodeExporter]

What steps will reproduce the problem?
1. run twitter XMPP transport via spectrum using microblog-purple 
(spectrum-1.4.1-1.el5, libpurple-2.6.6-5.el5_5, 
purple-microblog-debuginfo-0.3.0-3.el5 on RHEL 5.5/i686).
2. crash happens
3.

What is the expected output? What do you see instead?
no output, it should stay running
it crashes with this backtrace
Program terminated with signal 11, Segmentation fault.
#0  0x00d5808a in mb_conn_fetch_url_cb (url_data=0x94e6a68, 
    user_data=0x94e6920, url_text=0x0, len=0, 
    error_message=0x945d080 "Unable to connect to api.twitter.com: SSL Handshake Failed") at mb_net.c:165
165             retval = conn_data->handler(conn_data, conn_data->handler_data, 
error_message);
(gdb) thread apply all backtrace

Thread 1 (Thread 8145):
#0  0x00d5808a in mb_conn_fetch_url_cb (url_data=0x94e6a68, 
    user_data=0x94e6920, url_text=0x0, len=0, 
    error_message=0x945d080 "Unable to connect to api.twitter.com: SSL Handshake Failed") at mb_net.c:165
#1  0x058604bf in purple_util_fetch_url_error (gfud=0x94e6a68, 
    format=0x58b8a1e "Unable to connect to %s: %s") at util.c:3700
#2  0x05860c2b in ssl_url_fetch_error_cb (ssl_connection=0x94e69a0, 
    error=PURPLE_SSL_HANDSHAKE_FAILED, data=0x94e6a68) at util.c:4203
#3  0x06af957f in ssl_nss_handshake_cb (data=0x94e69a0, fd=6, 
    cond=PURPLE_INPUT_READ) at ssl-nss.c:340
#4  0x080cf44e in io_invoke(GIOChannel *, ._23, gpointer) (source=0x94e7d98, 
    condition=G_IO_IN, data=0x93e9ea0)
    at /usr/src/debug/spectrum-1.4.1/src/geventloop.cpp:51
#5  0x006afaad in ?? () from /lib/libglib-2.0.so.0
#6  0x00686372 in g_main_context_dispatch () from /lib/libglib-2.0.so.0
#7  0x00689366 in ?? () from /lib/libglib-2.0.so.0
#8  0x00689727 in g_main_loop_run () from /lib/libglib-2.0.so.0
#9  0x080dfb8d in GlooxMessageHandler::GlooxMessageHandler (this=0x92530e0, 
    config=...) at /usr/src/debug/spectrum-1.4.1/src/main.cpp:1065
#10 0x080e01fc in main (argc=153429452, argv=0x0)
    at /usr/src/debug/spectrum-1.4.1/src/main.cpp:2054
(gdb) 

What version of the product are you using? On what operating system?
see above

Please provide any additional information below.

Original issue reported on code.google.com by matej.c...@gmail.com on 15 Nov 2010 at 2:46

[GoogleCodeExporter]

Hi, 
thanks for the report.
Does this happen on connecting or on sending the message?
Do you know if spectrum has the capabilties to show the libpurple 
debug-messages / Can you attach debug messages from libpurple ?

Original comment by andresen.nils@gmail.com on 15 Feb 2011 at 3:50

• Changed state: Accepted

[GoogleCodeExporter]

When connecting to twitter, I haven't had a good Internet connection between 
the spectrum server and twitter.

Original comment by matej.c...@gmail.com on 15 Feb 2011 at 5:42

[GoogleCodeExporter]

Hi,
I found the libpurple-logs you provided in http://spectrum.im/issues/101
Could you please provide a recent log "containing" a crash ?

Original comment by andresen.nils@gmail.com on 16 Feb 2011 at 7:34

[GoogleCodeExporter]

It hasn't crashed for since then AFAIK

Original comment by matej.c...@gmail.com on 16 Feb 2011 at 8:49

[GoogleCodeExporter]

So, the crash has not happened since then and I am unable to reproduce it..
May I close this bug and you open a new one if this happens again?

Original comment by andresen.nils@gmail.com on 17 Feb 2011 at 7:40

[GoogleCodeExporter]

When you promise me that you will go through all code and check that you 
account for a possibility that network services are not available for any 
reason.

Original comment by matej.c...@gmail.com on 17 Feb 2011 at 8:16

[GoogleCodeExporter]

I could promise you to do that, but as I understand maybe a half of what is 
going on internally it won't do much good ;-)

But this gives me a new Idea: when you redirect api.twitter.com to an non 
existing server, say by editing /etc/hosts on the machine spectrum is running - 
that should simulate a "bad connection between spectrum and twitter", won't it? 
Can you reproduce the error that way?

(When I try the same in pidgin, I get an error, but no crash. Currently I am 
unable to test this in spectrum - But I am searching for space to install a 
virtual machine which can host a spectrum server now...)

Original comment by andresen.nils@gmail.com on 17 Feb 2011 at 10:34