cfd8d3f88d
Make PING skip the flood queue just like PONG does
...
It's unlikely server has nothing to send at all for all the other messages ZNC is sending and that there's nothing happening server-side too. But PING at least is guaranteed to give the response, so ZNC shouldn't disconnect too eagerly
2025-05-11 19:36:53 +01:00
aaeab22ea3
Fix processing of multi-token CAP ACK
...
I broke it in 5943bc9ed9#1950
2025-05-11 18:49:35 +01:00
cc1333fd5e
Update translations from Crowdin for bg_BG da_DK de_DE el_GR es_ES fr_FR id_ID it_IT nl_NL pl_PL pt_BR pt_PT ro_RO ru_RU tr_TR
2025-05-11 00:26:00 +00:00
4a15e1cf36
Fix unit test after previous change
...
See #1949
2025-05-10 21:00:22 +01:00
3d1551b7f2
Don't forward client JOINs during registration
...
ZNC remembers that it should join these channels, and will join them after registration. But if client automatically joins some channels, we don't want it to be added to send queue before parts of registration itself (CAP, AUTHENTICATE), because server will just disconnect with "Registration timeout". After registration is complete, using /join still joins the channel immediately.
Only limiting this to joins, because server may request some input from user to finish registration, and joins are the ones which are prone to be sent automatically by client to cause issues.
Fix #1949
2025-05-10 20:09:25 +01:00
8dab274ed8
Update translations from Crowdin for bg_BG da_DK de_DE el_GR es_ES fr_FR id_ID it_IT nl_NL pl_PL pt_BR pt_PT ro_RO ru_RU tr_TR
2025-05-10 00:25:54 +00:00
5e5d6cac59
merge
2025-05-09 23:59:54 +01:00
5943bc9ed9
Send several caps in the same CAP REQ
...
If they got NAKed, then retry them one by one
This should speed up the usual case
Fix #1950
2025-05-09 23:46:11 +01:00
e80b11b5b5
Move base64 table from header to .cpp
2025-05-09 21:54:36 +01:00
d707b41ac3
Add more stuff to .gitignore
2025-05-09 21:09:40 +01:00
2ca596ce8f
Update translations from Crowdin for bg_BG da_DK de_DE el_GR es_ES fr_FR id_ID it_IT nl_NL pl_PL pt_BR pt_PT ro_RO ru_RU tr_TR
2025-05-09 00:26:31 +00:00
c2f299abdf
Filter out server-dependent caps in CAP NEW using blacklist too
2025-05-08 23:25:58 +01:00
1c197a5508
Add a way to disable certain capabilities
...
This is a way for admins to mitigate some issues caused by caps if such issues ever arise.
E.g. add this to global level in znc.conf:
DisableClientCap = sasl
DisableServerCap = chghost
DisableServerCap = message-tags
Then these caps will be NAKed to client / not requested from server.
Note that this mechanism doesn't fully prevent a cap from being activated, e.g. one could use *send_raw module to request it from server even when disabled.
2025-05-08 21:55:40 +01:00
1063b7f5d6
CI: handle the case where junit is not outputted
...
though maybe github actions should also use one
2025-05-04 19:56:35 +01:00
ca6ae51179
CI: try to fix jenkins from counting every test twice
2025-05-04 19:41:11 +01:00
2f110896e7
CI: restore JUnit output processing in Jenkinsfile
...
via wrapper which modifies the environment per test
Workaround for https://github.com/google/gtest-parallel/pull/89
2025-05-04 18:38:43 +01:00
267b2fee99
ZNC 1.10.0-alpha1
2025-05-04 12:29:18 +01:00
553ce0bfd8
Mention boost version in readme
2025-05-04 12:12:02 +01:00
47c4fa6cd8
Merge pull request #1939 from DarthGandalf/boost
...
Find boost in CONFIG mode of CMake
2025-05-04 12:10:10 +01:00
50f085cc2f
Disable i18n test when i18n is disabled
2025-05-04 10:36:48 +01:00
1ec5408d7a
Find boost in CONFIG mode of CMake
...
To silence CMake's warning
Fix #1934
2025-05-04 10:29:50 +01:00
26d2b8fc86
Add doxygen comment to CMessage::GetType()
2025-05-02 19:32:30 +01:00
ca6d6cd27f
Update translations from Crowdin for bg_BG da_DK de_DE el_GR es_ES fr_FR id_ID it_IT nl_NL pl_PL pt_BR pt_PT ro_RO ru_RU tr_TR
2025-05-02 00:24:52 +00:00
5b871e74aa
Bring unix socket syntax of AddPort in line with AddServer
2025-05-02 01:00:56 +01:00
2a5dc04ef5
Merge pull request #1945 from DarthGandalf/unix
...
Unix sockets
2025-05-02 00:47:26 +01:00
00579d151e
Pass %p to LLVM_PROFILE_FILE instead of wrapper
2025-05-01 22:44:39 +01:00
2f02a9ea00
Merge branch 'master' into unix
2025-05-01 22:41:27 +01:00
92fc74cc00
Try to fix codecov config
...
Apparently they changed YAML parser, and ! now makes it fail
2025-05-01 22:41:13 +01:00
864252bfc6
update csocket
2025-05-01 22:31:12 +01:00
5974d0ff3b
Workaround for parallel writing of LLVM profiler data
...
Skip failing tests on cygwin: we'll just have to mark the unix socket feature as "experimental", but shouldn't block merging it due to cygwin being weird
2025-05-01 22:24:10 +01:00
056c7d8069
Explain how install ZNC in a custom path ( #1946 )
...
Added the the `-DCMAKE_INSTALL_PREFIX` for those wanting to install ZNC in a custom path.
Co-authored-by: Alexey Sokolov <alexey+github@asokolov.org >
2025-04-29 07:51:57 +01:00
5184d662d3
macos test probably should be fixed now
2025-04-28 22:08:13 +01:00
81e7fd69de
WIP debugging failures on mac and cygwin...
2025-04-28 00:55:13 +01:00
1ac3571c60
Add modified Qt to NOTICE
2025-04-22 01:15:23 +01:00
951c39479f
Update several tests to accept empty string or "localhost"
...
For unix sockets the behavior of getpeername() is different on different OS
This required to add support of regex to the integration test framework
2025-04-22 01:10:22 +01:00
d8234a1a71
Fix the IP in SASL
...
This is wrong PR for that, but I noticed the issue while debugging failed tests here
2025-04-22 00:10:21 +01:00
e084af4b7d
more hacks to the cygwin hack
2025-04-21 17:33:35 +01:00
174a57a6ab
fix typo
2025-04-21 16:38:33 +01:00
6b6614cffc
Adapt cygwin workaround to ancient cygwin's qt version
2025-04-21 16:36:39 +01:00
ce0a1584ba
Cygwin: try a horrible hack to fix integration test with unix sockets
2025-04-21 15:47:21 +01:00
2af8cef4c2
update csocket
2025-04-21 11:22:24 +01:00
fc15b8ec5e
Make 'make inttest' run in parallel even outside CI
2025-04-21 09:31:08 +01:00
4c8e780a60
Revert "CI: parallelize integration test"
...
This reverts commit b1a03fa1cc
2025-04-21 09:06:23 +01:00
81e3c908ad
Fix include
2025-04-21 01:01:42 +01:00
b1a03fa1cc
CI: parallelize integration test
...
I'm not sure whether 'make inttest' should do this or not
2025-04-21 00:52:29 +01:00
b642d92ce7
Switch integration test to mostly use unix sockets
...
By not using the same hardcoded number for every test, we can parallelize the test now.
There are several cases remaining where we can't easily use unix sockets (e.g. QSslSocket or imapauth module), for that ask kernel what port number is currently free to use. This is a bit racy though.
2025-04-21 00:19:46 +01:00
63d10ccb17
Add support to connect to server via unix socket
...
The syntax for AddServer command and config is chosen to be unix:/path or unix:ssl:/path
For security reasons, only admins can add such servers, to prevent users from poking around the file system.
2025-04-20 22:40:20 +01:00
66b17926cc
update csocket
2025-04-20 08:54:25 +01:00
d3a7f125cf
Make unix sockets work from modules
2025-04-20 08:31:36 +01:00
ea33a85ccb
Update translations from Crowdin for bg_BG da_DK de_DE el_GR es_ES fr_FR id_ID it_IT nl_NL pl_PL pt_BR pt_PT ro_RO ru_RU tr_TR
2025-04-20 00:26:50 +00:00
Alexey Sokolov