| Commit message (Collapse) | Author | Age | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes trac ticket #1248
proxy_connected() calls phb->func(), then tries to do phb_free() directly
afterwards, but that might have been freed by a proxy_disconnect() call
during the execution of that callback.
This one happened to several different people because some AIM server
broke recently.
This commit fixes it by implementing a phb_connected() function that
removes the PHB from the hash table before calling phb->func(), which
ensures that any proxy_disconnect() calls just close the fd and nothing
else.
|
| |
|
|
|
|
| |
Coverity says it could be an out of bounds read, but the value is set
internally, so not really. Still, good point about the condition being
wrong.
|
| |
|
|
|
|
|
|
|
|
|
| |
Fixes trac ticket 1198, https://bugs.bitlbee.org/bitlbee/ticket/1198
This function can be used as a safe drop-in replacement to closesocket()
If a proxy connection is pending (connected callback still not called),
it looks up the PHB in a hash table indexed by fd. If it is there, it
closes, frees the phb and avoids further calls to the callback.
If it is not in there, it just does closesocket()
|
| |
|
|
| |
More cleanup.
|
| |
|
|
| |
Just cleanup.
|
| |
|
|
|
|
|
|
|
| |
Fixes trac ticket 995 https://bugs.bitlbee.org/bitlbee/ticket/995
This is slightly pointless for the suggested use case (tor), since with
socks5 we already send a hostname instead of an IP address.
Either way, it was easy to implement, so I hope it helps.
|
| | |
|
| |
|
|
|
|
|
| |
Used uncrustify, with the configuration file in ./doc/uncrustify.cfg
Commit author set to "Indent <please@skip.me>" so that it's easier to
skip while doing git blame.
|
| |
|
|
|
| |
It was checking for "Connection established" and some proxies use a
different string, such as "Tunnel established" in polipo
|
| |
|
|
| |
Both in gaim_io_connected(), visible when using an http proxy
|
| |
|
|
|
|
|
|
| |
* Don't use PIE/PIC on Cygwin/Darwin unless specified as these
platforms don't support it.
* Cleanup warnings for 'make check' build.
* Fix the type issue for getsockopt calls.
* Fix enum warnings in Yahoo libs on Mac OS X.
|
| | |
|
| |
|
|
| |
It has been broken for a very long time and nobody cared about it.
|
| |
|
|
|
|
| |
let me close #641, and more importantly, prepares the Twitter module for
an upcoming API change. https://dev.twitter.com/blog/deprecating-http-1.0-streaming-api
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
This should fix issues with hosts that have IPv6 and IPv4 addresses but listen
on only one of them. (Bug #673)
This also fixes a bug that broke error checking in gaim_io_connected(), until
now event handlers were never actually getting proper error reporting (fd=-1),
but IIRC they should all handle it anyway as I was never aware of this bug.
|
| |
|
|
|
| |
lets proxy_connect() connect to IPv6 hosts.
|
| |
|
|
|
|
|
|
|
|
| |
diverged too much this is fairly transparent. I did rename and redefine
GAIM_INPUT_* variables to really make it work without adding another stupid
layer in between.
One problem left, the new libpurple input API doesn't care about return
values. Fixing that in the next CL.
|
| |
|
|
|
| |
invoked but was stupid anyway). #409.
|
| |
|
|
|
| |
specific interface before connecting to a remote host.
|
| |
|
|
|
|
|
| |
to Miles Bader for reporting this in the Debian BTS. Apparently not many
people use this functionality, it was broken in bzr for more than a year
already...
|
| |
|
|
|
|
| |
a non-blocking connect() can return immediately (when connecting to
localhost, for example). Closes bug #233 and #340.
|
| | |
|
| | |
|
| | |
|
|
|
protocols/, and adding RC4 code.
|
