adf56ad8fd
Re-enable FreeBSD workaround for mysterious test failures.
2019-06-07 12:24:25 +02:00
e423f69082
[tkldb] Add some extra check for trailing DB data.
...
If everything goes correctly then after reading all TKL entries we
should be at the end of file. If there is still data after that,
something went wrong... quite wrong.. :D
2019-06-05 09:39:59 +02:00
673779eab9
Write string lengths as 16-bit rather than 64-bit: saves about 34% space
...
of the database file, with possible some small speed improvement as well.
2019-06-05 09:35:43 +02:00
ec5811dc3b
[tkldb] Save every 5 minutes now that writing is fast enough :)
...
And mention the benchmark figures in the .c file.
2019-06-05 09:01:04 +02:00
eaa1b213b9
[tkldb] Use buffered I/O. Writing the DB is now 29 times faster.
...
Saving 100k zlines now takes 72ms instead of 2100ms
Loading 100k zlines now takes 510ms instead of 1300ms
2019-06-05 08:53:30 +02:00
1f6bb6d3cc
Some benchmarking
2019-06-05 08:45:48 +02:00
498f728cbb
[tkldb] If DB is corrupt then rename to .corrupt and start a new one.
...
This, rather than having the module not loaded at all, which could mean,
especially if missed the warning on boot, that you run for weeks or
months without having your TKL's stored, which would be a shame ;)
Also a failure to rename() is not fatal, as it likely means that we
don't have permissions, in such a case you will see a repeated error
every X minutes due to the write, which is good.
2019-06-05 08:22:22 +02:00
21f24cc596
Fix for NICK+USER order. Should be correct? Should check identd..
2019-06-04 20:05:27 +02:00
5f8dd67051
if( -> if ( :D
...
[skip ci]
2019-06-04 20:04:22 +02:00
e9da2eb3cb
[tkldb] "skipped".. why? "expired" :D
...
And don't bother users too much about this, unless in DEBUGMODE.
2019-06-04 20:00:15 +02:00
6b1af917fc
Do it the other way around than previous commit. Force 32 bit type.
...
So it's compatible with machines where int is 64bit (ILP64),
mostly for the future I guess.
2019-06-04 19:52:59 +02:00
628f367849
tkldb: version field should also be forced 64 bit or you'll have the same
...
problem with porting 32 to 64 again...
2019-06-04 19:34:37 +02:00
52b39acf78
Fixed return values/checks for helper functions/macros, return empty string instead of NULL when reading zero length strings, discard tkl->subtype and store spamf as a char instead of unsigned int, remove rewriting after DB read, also renamed TKLines/TKLs/etc to *-Lines for uniformity with other code
2019-06-03 20:45:56 +02:00
bb91ac1c56
Update this a bit. Will take a look again later.
...
[skip ci]
2019-06-03 18:48:18 +02:00
9f64711113
Remove FreeBSD 12 workaround
2019-06-03 15:17:02 +02:00
597186d9c3
BuildBot: Fix run-tests on FreeBSD 12
2019-06-03 14:45:56 +02:00
253df2c8db
BuildBot: -j3 -> -j4
...
[skip ci]
2019-06-03 13:31:58 +02:00
55757a5605
Remove references to old test framework.
...
[skip ci]
2019-06-03 13:30:15 +02:00
73cea64c9f
Updates for IRC test framework: changed port to 5901 of irc1
2019-06-03 13:03:43 +02:00
bc8e4a5c47
Updates for new IRC test framework
2019-06-03 12:59:22 +02:00
321cf75fb7
Resolve resolve merge conflict :D
...
[skip ci]
2019-06-01 08:57:19 +02:00
13f96b0a57
Small adjustments to tkldb: convert some to safefree(),
...
get rid of sizeof(char) - which is guaranteed to be 1 by C99 and
on any decent compiler and always use MyMallocEx even when it's not
necessary :D.
2019-06-01 08:53:17 +02:00
e78e303314
Re-indent tkldb, sorry could not resist :D
2019-06-01 08:46:22 +02:00
2556ae9d70
[tkldb] Allow boot without existing database. Change return values to 1/0 (true/false).
2019-06-01 08:39:32 +02:00
b8e65a2f7f
Buildbot: *NIX build test: add -DNOREMOVETMP to make valgrind do a better
...
job at printing module information (I hope).
2019-06-01 08:24:31 +02:00
7c32e4d63f
Windows build: fix makefile.win32
2019-06-01 08:23:33 +02:00
4dbe8a8238
Resolve merge conflict :D
2019-05-31 22:39:27 +02:00
c697e2ea99
Renamed m_storetkl to just tkldb
2019-05-31 22:25:18 +02:00
bf6f885b98
Storetkl now also supports the hashed *@IP Z:Line thingy, also added errno calls to all error output
2019-05-31 22:06:59 +02:00
c43a1b15af
Store TS types in a uint64_t before DB writing to prepare the module for a Year 2038 fix, also upgraded num/rewrite ints to 64 bit since they *might* be as big as tklcount
2019-05-31 21:23:19 +02:00
c6389270f9
Remove a few useless #define's (And final test of buildbot today :D)
2019-05-30 21:02:27 +02:00
c3add3b41c
Windows build: Randomly remove some slashes, for no reason.
...
[skip ci]
2019-05-30 20:12:25 +02:00
7d7d736df3
More windows build fixes
...
[skip ci]
2019-05-30 20:03:50 +02:00
858b63e1d5
Windows: remove .CHANGES.NEW
...
[skip ci]
2019-05-30 19:56:54 +02:00
9af24a56f9
Buildbot: makefile.win32 updates
...
[skip ci]
2019-05-30 19:42:17 +02:00
57e06a7070
Windows: add api-mtag to makefile.win32
2019-05-30 19:36:15 +02:00
f0a773ee59
Windows: Add conf_preprocessor to makefile.win32
2019-05-30 19:33:21 +02:00
a2a7acc4bd
Buildbot: more windows updates
2019-05-30 19:30:20 +02:00
e8e113a8c0
Buildbot: .............................
2019-05-30 19:23:15 +02:00
120ec5cea3
.............
2019-05-30 19:22:14 +02:00
d02b4de5d3
Buildbot: updates for windows
2019-05-30 19:20:15 +02:00
2682d7bc6d
Buildbot: Updates for windows
2019-05-30 19:08:33 +02:00
9530006e70
BB: Terminate unrealircd at end of test (otherwise it may hang buildbot)
2019-05-30 16:48:05 +02:00
03fbc35fb7
Buildbot: Run cipherscan not on FreeBSD for now, so we can test the rest.
2019-05-30 16:14:06 +02:00
3f723323d0
BuildBot: Use gmake on FreeBSD and temporarily disable select-config.
2019-05-30 16:08:54 +02:00
ecac3f9d4a
Mention U5 in ./Config -quick restriction
2019-05-30 08:52:08 +02:00
1b0a7cb9b8
Fix build tests due to previous change.
2019-05-30 08:51:05 +02:00
195cd17d02
Rename LIBDIR to PRIVATELIBDIR in config.settings to avoid confusion.
2019-05-30 08:46:13 +02:00
4004152a7a
Updated m_storetkl's descriptions for added clarity
2019-05-28 21:44:27 +02:00
0f97c03487
Probably a good idea to add storetkl stuff to Windows makefile too huh
2019-05-28 21:38:38 +02:00
Bram Matthys