It turns out that it's possible to receive data before an upload
finished when the server sends HTTP_BAD_REQUEST in the middle of the
upload (this happens to me when I try to upload a 6000x6000 image to my
profile feed, with is a 44MB file). So, in that case the finished upload
detection did not fail and we shouldn't assert.
Introduced in
6dcda3595e
(Add recovery for randomly closed socket desciptors.)
By copying CurlSocketInfo* into mCopiedFileDescriptors, it was possible
that we accessed a deleted CurlSocketInfo for it's filedescriptor,
returning a random value, which, when passed to FD_ISSET could cause a
SIGSEGV.
I reverted this change and now store the literal socket descriptors in
mCopiedFileDescriptors again. It is not a problem when libcurl deletes
the corresponding CurlSocketInfo (through a callback to
MultiHandle::socket_callback) in that case, because the fd_set we test
against isn't updated as a result of that, not until we're down with
mCopiedFileDescriptors (at the re-entry of select()).
* Split off AIThreadSafeBitsPOD, because offsetof may only be used on POD types.
* Added ASSERT_ONLY and ASSERT_ONLY_COMMA
* Removed a few unused class members
* Fixed a bug in AIHTTPReceivedHeaders::equal that more or less only
compared the length of the headers before :/
Although it should never happen that a file descriptor is suddenly
closed, it appeared that this happens on linux 64bit when using
FMODex... Not really sure how useful this is, but at least now the
viewer just continues to work, as if -say- the socket was closed
remotely. Before the curl thread would go into a tight loop that it
wouldn't recover from until the watchdog thread terminated the viewer.
When linking against the prebuilt libcrypto.so.1.0.0 and
libssl.so.1.0.0 there is a problem running the viewer in
gdb (on debian) because gdb links with libpython2.7.so
which needs a newer openssl on my system (OPENSSL_1.0.1).
The patch allows to work around this problem by defining
-DUSE_SYSTEM_OPENSSL:BOOL=ON when configuring, which then
causes the viewer to be compiled against the system libs.
(If you already installed the prebuilt, you have to manually
remove them with script/install.py --uninstall openSSL).
The prebuilt libcurl expects a function SSLv2_client_method
however, which is not present in my openssl system libs.
A stub was added which is possible because the function
in question isn't used anyway.
CURLE_COULDNT_CONNECT only happens when we actually get
an error (ie, 'connection refused'). If we time out while the connect
attempt is still running then we get CURLE_OPERATION_TIMEDOUT.
Cookies are now collected during redirects, one the last cookie
of a given name is kept. Cookies are then set by looking for
the right cookie name (instead of what viewer 3 does: just
use the last cookie that was set).
This fixes the merchant outbox.
Add CURLTR debug channel for libcurl API calls,
and use CURLIO only for libcurl debug output.
Note: need to set gDebugCurlTerse to true for
filtering to take effect, then pass 'debug_on'
to the LLHttpClient methods that require debugging.
Conflicts:
indra/newview/llfloateravatarpicker.cpp - Includes to v-d style, since plain alphabetical just won't cut it. Shyotl's changes for name system, here.
indra/newview/llpanelgroupinvite.cpp - Shyotl's changes for name system, here.
Also corrected new README to have less typos and be slightly more accurate.
Better viewed without space changes
Pretty much add support for sending LLAvatarName and std::string to become a name via the system...
Also a bit of a revert for llcallingcard.cpp in this regard, just allow it to use a callback along with the new support,
but still handle the actual name switching in LLAvatarNameCache.
Fixes a problem I was having with "[NAME] is Online" notifications appearing early on, only.
Note: Large groups still don't load?!
SimConsole cap is needed for Sim Console, won't work without it, keeping it in!
..But it should work with the new ASync one? Maybe needs tweaks, first?
Also switches opening IMs to the gCacheName->getFullName(id, name) system mentioned earlier, my connection has improved!
And makes the Avatar Picker show display names! Yay! (Rearranged the includes there, alphabetical~)
Also fallback to gCacheName->getFullName(id, name) if getting name fails in the new system.. hopefully this will combat occasional big failures.. but it may not work anyway.. maybe my connection is just awful again
Only does that, dun believe me? Take a look, oh... it also corrects improper interpretations of the PhoenixNameSystem value: 1 is both, 2 is displays, 0 and anything else is Legacy.
Also replaces a bit of the display name support code with new system... by definition this shouldn't change the log file, only the title of the session..
because of my bad connection during week days, I can not confirm this working 100%,
but if it does end up changing the logs under certain unseen conditions, replace the new name check with gCacheName->getFullName(id, name) but this reallllly shouldn't happen!
Before every HEAD and GET request allowed redirection by default,
without setting a limit on the number of redirections. This caused
an infinite redirect loop when connecting to marketplace, in combination
with the bug that we did not allow cookies.
