| Branch | Commit message | Author | Age |
|---|
| amdj/autotools-version-fix | include/serno.h.SH: correct check for git directory |  Aaron Jones | 3 months |
| amdj/fix-buildsys-headers | Clean up include/stdinc.h and librb/include/rb_lib.h | Aaron Jones | 3 months |
| amdj/list-memleak | Merge branch 'main' into amdj/list-memleak |  Doug Freed | 7 weeks |
| amdj/meson-generation-builddate | Meson: Decouple creation and generation data from version | Aaron Jones | 4 weeks |
| amdj/modules | Tidy up module loading code | Aaron Jones | 5 months |
| amdj/openssl-strerror | OpenSSL backend: Use nicer error strings for system errors | Aaron Jones | 4 months |
| amdj/unify-openssl-macro | Unify the preprocessor macro for OpenSSL support | Aaron Jones | 5 months |
| dwfreed/meson-rebased | build: add Meson buildsystem |  Valentin Lorentz | 5 months |
| dwfreed/rm-rf-autotools | build: remove autotools buildsystem | Doug Freed | 4 months |
| main | Only init solanum once in tests/sasl_abort1.c |  Ryan Schmidt | 13 days |
| [...] |
| |
| |
| Age | Commit message | Author | Files | Lines |
|---|
| 13 days | Only init solanum once in tests/sasl_abort1.c•••Calling solanum_main multiple times results in loading every
core/autoload module multiple times (because the check for "is the
module already loaded" always returns "no" due to solanum_main clearing
the list of loaded modules before attempting the load). Init only once,
and use other code to ensure that UIDs are consistent instead by
resetting the counter and ensuring we free the clients with the old UIDs
at the end of each test run.
HEAD main | Ryan Schmidt | 1 | -11/+9 |
| 13 days | Call the channel_lowerts hook on SJOIN too•••Previously this hook was called only when a remote join outside of a
netjoin burst to an existing channel happens, which is a fairly odd set
of criteria. By making it run for SJOIN too, it now covers every case
where a channel's TS is lowered, making the hook more reliable for use.
| Ryan Schmidt | 2 | -11/+15 |
| 13 days | Add user_welcome hook•••This hook is executed when the registration burst is being sent to a new
local user, after sending RPL_ISUPPORT but before sending LUSERS output.
Many IRCv3 specifications use this timing as the point to send
additional information during registration bursts, so having an easy
hook point to do that makes implementing such specs possible.
| Ryan Schmidt | 4 | -1/+16 |
| 13 days | Expand doing_whois/doing_whois_global hook data•••Use hook_data_client_approval instead of hook_data_client so that we can
pass whether operspy WHOIS is being performed as part of the hook data.
hook_data_client_approval is backwards compatible by having the same
beginning layout as hook_data_client, so modules compiled against the
old hook data should have no issues.
Unlike most hooks where an approval struct is passed, the approved field
is meant to be read-only here. The approval struct was simply used out
of convenience vs having to define a custom struct with identical layout
for the case of WHOIS.
| Ryan Schmidt | 2 | -5/+10 |
| 13 days | Add sendto_list_local_* and sendto_monitor_with_capability•••These are needed for METADATA support, and sendto_monitor was refactored
into a common sendto_list_local_internal method used by all of the
newly-added functions as well. Test cases were also added to excercise
the new functions.
| Ryan Schmidt | 3 | -17/+432 |
| 13 days | Expand monitor•••- new is_monitoring function that modules can use to determine if a
particular client is monitoring a particular nickname
- new hook that is fired when a user adds a nick to their monitor list
| Ryan Schmidt | 4 | -1/+46 |
| 13 days | Make change_isupport useful•••The function now accepts an optional pointer argument to a new struct
that contains all of the data of the previous isupport function (both
callback and data), rather than only returning the previous data
pointer. This in turn allows modules to both call this function to
overwrite an isupport handler in modinit as well as restore the previous
handler in moddeinit (something that was impossible before).
The return value was changed to int, returning 0 if the named isupport
method didn't previously exist and 1 if it did exist, so the caller can
judge whether this function actually accomplished anything.
There were no callers of this function due to it previously being of
questionable use, so no other code needed adjustment.
| Ryan Schmidt | 2 | -8/+20 |
| 13 days | Pass block name to conf_start callbacks•••Configuration blocks of the form block "name" { ... } previously were
unsupported in extensions calling add_top_conf because the name portion
was not exposed to anything within the extension. Pass the name along so
callbacks can use that data instead of relying on global state in
newconf.c with static storage duration.
| Ryan Schmidt | 2 | -17/+37 |
| 2026-07-19 | Meson: Decouple creation and generation data from version•••This fixes the generation counter being hardcoded at 0 and the
IRCd birth date being reported as "__DATE__ at __TIME__".
| Aaron Jones | 13 | -83/+114 |
| 2026-07-14 | Move setup.h include above rb_lib.h and fix meson assertion printout•••In autotools, NDEBUG is defined in setup.h, which means we need this included before we include any system headers which depend on it (i.e. assert.h). As such, make this the first include in stdinc.h instead of transitively including it from ircd_defs.h.
Also fix the assertions printout provided to the user as b_ndebug=true disables assertions, rather than enabling them. It also takes an if-release value, so detect that properly too.
| Ryan Schmidt | 3 | -3/+9 |
| [...] |
| |
| Clone |
|---|
| https://cgit.space/~irc/github.com/solanum-ircd/solanum.git |
| http://cgitspacea7m5fmo5stfwxkykfbrjcecevl6z3xm5u7nfppuiigqoeyd.onion/~irc/github.com/solanum-ircd/solanum.git |
| http://cgit.i2p/~irc/github.com/solanum-ircd/solanum.git |
| http://cgitiek6febqkrplowpeqssm6cur3fk6de76oajer7dqixw2pwkq.b32.i2p/~irc/github.com/solanum-ircd/solanum.git |
Aaron Jones
Doug Freed
Valentin Lorentz
Ryan Schmidt