| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
| |
Somehow I'm not sure whether this thing is useful in any sense.
|
| |
|
| |
|
|
|
|
|
|
| |
`-Weverything' seems to have found a few problems.
Also enabled clang sanitizers by default.
|
|
|
|
|
|
| |
To save us from specifying the `next' and `prev' pointers all the time.
It's not perfect, just a tiny bit better.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
We have to release the old nickname _first_, so that we can register the
same one again (albeit with a slight race condition).
|
| |
|
| |
|
|
|
|
| |
Now we're a real daemon, yay.
|
|
|
|
|
|
| |
Let's limit print_fatal() to unexpected conditions.
Also added exit_fatal() to save a few lines of code.
|
|
|
|
| |
Otherwise TLS autodetection and TLS communication itself seem to work, yay!
|
|
|
|
|
|
| |
It seemed to be almost too easy to be true, and it actually was.
Now it should work.
|
|
|
|
| |
Now we can at least successfully register.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As it turns out, they're rather annoying to maintain, and we don't even need
them. They also clutter the code unnecessarily in their current form.
If it ever comes to having to have them, let's make another version of
error_set(), maybe error_set_with_code(), that makes it possible to also set
an integer within `struct error'.
The only problem with the above solution is when we aggregate errors from
multiple functions (be it by calling one after another, or through nesting of
functions that may return an error).
But let's care about that when the time comes for it.
|
| |
|
|
|
|
| |
This should make the programs an awful lot less painful to set up.
|
|
|
|
|
| |
It is shorter and makes more sense. Also replaced "conn" with "c"
to keep it even more concise, and thus clearer.
|
|
|
|
|
| |
And shuffle around some functions so that they form logical blocks (at least
I've tried; it's not that easy when you try to avoid forward declarations).
|
| |
|
| |
|
|
|
|
| |
Later I'm going to put in there some more information about the format etc.
|
|
|
|
|
|
|
| |
This is going to enable making changes to ERR and RPL messages without
modifying the source code.
Localized messages could be interesting. :)
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
It didn't make much sense to parse the configuration values and load the SSL
keys on each connection.
|
|
As it is going to share a great lot of functionality with ZyklonB, I have
decided to extract the common parts into `common.c' and make the two subprojects
include this file.
The Single Compile Unit concept has proven valuable (sub-second compile times,
dead code warnings, almost no need for function declarations, whole-program
optimizations), and the sources aren't that long so far anyway.
I am probably going to add CMake support later but so far it's not a priority.
This project is meant to be portable and freely relocatable (ie. no hardcoded
paths if I can avoid it).
|