reduf / headquarter Goto Github PK

Compiling on Windows (can we get a label for this?)

Had a crash when subscribing/unsubscribing to an event via events_add_entry and events_rem_entry in events.c, see below

void broadcast_event(EventManager *mgr, EventType e, void *args)
{
    assert(0 <= e && e < N_EVENT);
    CallbackEntry *entry;
    struct list   *it;
    list_for_each(&mgr->callbacks[e], it) {
        entry = list_entry(it, CallbackEntry, node);
        assert(entry); // This should never happen if `events_rem_entry` is working correctly?
        if (entry->callback) {
            entry->callback(e, args, entry->param);
        }
    }
}

I've added the assertion here to flag it up; in theory entry should never be a null pointer, is this correct? Otherwise we need

        if (entry && entry->callback) {
            entry->callback(e, args, entry->param);
        }

To catch it, but this feels wrong too.

The offset of the message is wrong, because it was increased in a game update.

Heya I was wondering for which version of the game this tool is?

Because it seems pretty outdated, but already got the new DH Params, and created a new patcher for it (using probabilities), new packet handler (supports Lz4, and multi/split-packets), and got the packets extracted by modding the game, which hooks into the registration functions before the initialization.

We read the otp code as an integer and subsequently serialize it as an integer. This mean that if a code starts with 0(s), we loose the 0(s).

We could instead keep a string and simply forward user input.

In the following example, field.size is 0. This is a problem when deserializing the structures.

init_player(player); needs to be called somewhere inside HandleAgentCreatePlayer before copying the name from the packet

There was one instance of %S logging that was wrong here: https://github.com/reduf/Headquarter/blob/e200a554135235f2f72dca8e95837a18b653ef58/code/client/guild.c#L174L180

We have to be careful, because on Linux wchar_t are usually 4 bytes and that would crash. Also, I'm not sure we can assume and I don't think we wants to assume the nul-termination here.

When compiling a plugin with client as a dependency, CMAKE_SOURCE_DIR isn't the client source dir - use CMAKE_CURRENT_SOURCE_DIR as per common and portal CMakeLists files

Change function definition from malloc to calloc to avoid uninitialised struct vars

static void ensure_agent_exist(GwClient *client, AgentId id)
{
    ArrayAgent *agents = &client->world.agents;
    if (array_inside(agents, id) && array_at(agents, id))
        return;

    if (!array_inside(agents, id)) {
        array_resize(agents, id + 1);
        agents->size = agents->capacity;
    }

    Agent *agent = calloc(1,sizeof(*agent));
    agent->agent_id = id;
    agent->speed_modifier = 1.f;
    array_set(agents, id, agent);
}

The player array is bigger than the number of players, because the index are the player number. It's not necessarly a problem, but GetPlayers is defined as:

ArrayPlayer players = client->world.players;
if (buffer == NULL) {
    count = players.size;
    goto leave;
}

So, we could keep track of the real number of players.

Probably an error in array_push.

Seems to crash inside appendf via portal login

    appendf(&content, "<ConnType>400</ConnType>\n");
    appendf(&content, "<Address>%s</Address>\n", sockname); // Segmentation fault here

No issues with windows

Assertion fails without it created