Are you using the CMake build system? In that case, you need to enable the server building either using ccmake or similar, or setting the flags on command line:
cmake -DBUILD_DEDICATED_SERVER=YES -DBUILD_MASTER_SERVER=YES ..
After which the servers will be built as separate binaries with plain make -j3 without extra parameters.

This is very nice: it now took something like 15 seconds to clone the entire repo (yes, I do have a rather speedy connection), while previously that wasn't enough time for checkingout one svn revision. Nice to also know there has been development, I was a bit scared when the svn had not received updates. My sf.net id is aave000. I guess I should find out where I put my idea todo.

Hi,
Somehow I don't get errors, but I added OpenGL to the CMake. Commit:
https://github.com/tapio/coldest/commit/f1235f7cf36d060a3a5141ba62200a5a14aac7c1 "Add OpenGL to CMake."

The reason why simply adding it to the list won't work, is because the variables are a bit of a mess. I.e. I like to just loop over the libs like I do there (keeping the CMakeLists.txt short), but the variable names in the cmake find scripts are not consistent or are deprecated, so I need to modify them and put a script to the cmake folder.

EDIT: As I don't get errors, I don't know if this works but I hope so.

Actually for lone developer distributed system is better from the start: since he doesn't need to synchronize stuff with others (apart from uploading stuff every once in a while), the absence of network latency with Git and friends is a huge time saver.

Anyway, I wouldn't recommend sourceforge. It has slow ssh (not to mention the really slow "hosted apps" like trac), ugly web interface and is a bit of a dinosaur in many aspects. I personally like Github.com: it has prettier interface and the whole social aspect together with the possibility to easily use "fork -> hack -> make a pull request" workflow together with branch network graphs.

One slight problem with github is that is has a quota of 300 megabytes (which is about the size of the entire coldest svn repo history converted to git). However, that quota is actually a soft limit put in place to avoid abuse: it is not enforced in any way and the staff have said in numerous support threads that it's ok to have bigger, even 1Gb, but for best operation, should stick to 500M per repo.

Another rather popular git hosting site is Gitorious.com, but I don't have much experiences with it. I do think that Github offers more features though.

EDIT: To be fair towards sourceforge, it does seem to offer more detailed repository access etc. statistics if those are really important.

Okay, after many tries I managed to download the SVN repo and as promised the initial version of the CMake build system can be found from my hacking repo @ github: https://github.com/tapio/coldest
More specifically, the diffs of the commit-in-question are here: https://github.com/tapio/coldest/commit/f2c51530c33a62e18b68b99d4bec7c8fe1b4b2a8

Comments and testing experiences are welcomed.

I have now a working CMake build. It has switches for building each of the client, dedicated and master server (can be built in any combination e.g. all at once). There is not yet an install target because in order to be unix friendly, I think the resource thingy would need to be patched to understand the unix hierarchy ($PREFIX/bin for binaries, $PREFIX/share/games/coldest for data).

Once the SVN server is up and running again, I'll create a Git mirror for myself (because git is just so much faster, better and prettier than svn) and put the thing online for review. If you like it, you can then give me commit access and I can push the stuff to the official repo (I use git-svn two directional bridge for that - you wouldn't even know I actually use git ). Of course you can just apply patches if you wan't to keep the repo access to yourself, but in my experience giving access away easily is the most hassle-free way for everyone and encourages further contributions.

...And it's down again. Several days now I think.

EDIT: I'll take this time here to bash SVN a bit. This situation is one of the reasons why distributed version control systems (such as Git and Mercurial) are superior to centralized ones. When a distributed system's "central" repo goes offline I can still commit stuff, examine history, do branching and merging and everything in my local playground. When the official repo comes back online, I can just push my changes like nothing was wrong. Here with SVN I have a commit ready, but since I cannot commit it, I really don't want to code more since that would make it harder to create atomic commits later.

Hi,
I just tried Coldest today (compiled from svn on linux) and was rather positively surprised about it. I especially liked how most of the graphics didn't actually make my eyes bleed (apart from the hit effect stuff which is rather horrible) and how you've resisted xml/json whatever in favor of simple plain text config files (e.g. maps seem to be easy and nice to create apart from the procedural trees for which I would suggest a similar placement method as grass has).

Now I'd like to ask some questions.

1. Is there an IRC channel for this project?
I feel it's easier to discuss stuff and ask quick questions in real time. Also, if the project should have more than one developer, a CIA.vc IRC bot announcing all commits is pretty much de-facto standard for open-source projects.

2. Is the engine completely home-brew or is it based on something?
My quick initial analysis suggests it's original. It seems nicely not-ancient (using VBOs, FBOs, shaders etc. contrary to most open-source games sticking to OpenGL 1.4 and fixed function pipeline).

3. Are the bots completely brain dead and invulnerable?
As it seemed when I tested.

4. If I would construct a CMake-based build system, what are its chances of getting included or even replacing the current one?
I've grown rather fond of this popular cross-platform build tool that generates (among other things) colorful makefiles and most importantly keeps the source tree clean of the object code pollution and other compilation garbage. I actually almost completed this for my own use already, but encountered a small obstacle at the end when I realized I cannot simply glob sources with wildcards (*.cpp) as the server and client files are in the same directory.

Oh, I could tell you that I consider myself quite good C++ programmer with experience on OpenGL usage also. If I can find the time, I could consider some contributing.