TheLuda

All I can say for now: interesting recommendation. Worth a look.

There will be no 3.0.2 branch. 3.0.3 client is what we are heading for.

I am not sure if that properly identifies this as a development branch :cool:

Here comes the branch. Cleverly named 303-willcrashforsure and pushed to GitHub. For now it is empty or better said it has no changes. If you run a git pull you should get that branch on your system. Running git checkout 303-willcrashforsure will switch your local repository to the new branch. Please try to make all changes in small (atomic) commits, which represent logical units. E.g. one commit for all changed enums, one for changed opcodes, etc.

It's probably the best, to stick with 3.0.3 patch instead of 3.0.2 client.

There are some technical issues with the process. It's still work in progress.

Yes, it is. Plain git mirror over there with autoudate.

Our primary focus currently is not client updates, and not specific game features. It is more on core components of the server itself, like handling huge amounts of network connections, or having improved database connections.

Git is made for atomic commits, which means you should reduce patches to the smallest possible unit. It is better to have separate branches for each change.

I like it, Eliza returns and once again plays AI games

Git is the tool, to make addition of patches work better. Nothing else to add. We will ban nobody for a different opinion. Get used to Git, learn how to fork, create branches, and stick to the Coding Standards described on both the GitHub Wiki and the Lighthouse issue tracker, and you will see in increase in the speed of patch inclusion.

If you develop an installer for Windows, please use Windows Installer and not NSIS. Windows installer supports everything needed, including verification of dependencies, merge modules for required libraries, it can create update packages - and those are needed if we want steady rebuilds for each commit - and it is supported inside Visual Studio. Additional benefits of WiX include multi-language support, and easy installer creation from Visual Studio, as there is a plugin. Please do not use non-standard installers like NSIS, or InstallShield. They all in some points violate the Windows rules for application deployment. I would prefer a clean installer more, even if it takes longer to create.

Luabind might be a good approach to develop LUA bindings for MaNGOS.

Yes, this is worth the effort. Suggestion: check out the MySQL server setup repository (Subversion repository url). The installer is written using the Windows Installer Toolkit aka. WiX. Recent release are updated weekly, and a pretty much complete tutorial for WiX is available, too. The MySQL server setup is a pretty good template for creating a Windows installer for MaNGOS. Have a look at the repository, and enjoy the fun of Windows installer.

And b) has the chance of getting people educated That is if you believe our users can still be educated. I would still go the public route and add a startup warning to the demons along the lines of: Or we simply name the branch 302-willcrashforfree.

This sounds very much like adding the client tools, and the tools to develop actual data to MaNGOS, and turn it into a whole MMORPG eco system. The idea itself is not bad. But it surely will come with some development costs, and would require a very well thought out development workflow. Doable? Yes. There just needs to be an aim inside all that. And it has to be a very good one, because if you want to target mass market, you can see that besides Blizzard Entertainment other MMORPG developers hardly see the ground when it comes to attracting the mass market.

It will surely result in spam for 3.0.2. But it does not really matter, does it? With every new client there have been preliminary test patches, and even though they where published unsupported, people demanded support. Maybe doing this in the public and demonstrating the work behind a client update may change some peoples attitude to be less annoying about upgrades for the future. Well, maybe that will not happen. But we have not tried that yet, and probably doing this in the open might help. And if it does not change a thing, we will still end up with the same situation. If we create client updates in private, and then suddenly switch, we still end up having update spam in the meantime. I'd just hope for some positive change, when doing upgrades in the public. If it does not work, there is not much lost. I would not enforce the public upgrade, I just think it might make a difference in the peoples attitude and give us the chance of some real testing. As expressed in some other topics, there are people out there interested in the upgrade process, and some even might be willing to help or at least provide valuable feedback.

You will get your chance to see, what it takes to update the servers to support a client with updated object structures, and data packets changed. This comes with a price tag. The 3.0.2 branch will be completely unsupported and only posts with a clear indication of development interest will be replied. There will be no installation support, nothing.

Subversion isn't user-friendly either, but it looks sexy with TortoiseSVN.

A free, and well readable tutorial is available on http://www.cplusplus.com/doc/tutorial/ A free C++ reference (maintained in a Wiki) is available here: http://www.cppreference.com/ Both are good starting points, and the C++ reference is available for download, too.

Another update: I now have tried several approaches in updating the SVN repository, and there are a few issues left to fix. I guess it will take another day or two till I can fix all bugs in the conversion script.

Yes, it is a big upgrade. 3.0.2 changes a lot of core features, and that is a lot of work. We will probably create a Git branch for the 3.0.2 updates, in order to update the core. But that branch will be unstable for sure quite some time.

I am currently testing the conversion on svn://svn.mangosproject.org/MaNGOS/. Press thumbs. If everything works, all Git commits should be converted and end up in svn://svn.mangosproject.org/MaNGOS/trunk/ More updates coming soon.

It is no diss, he is referring to real things that happened, and that annoyed other developers, too. Wyk3d is just kind enough to not go into further details

Just to give you an impression: Patch 3.0.2 requires to make bigger changes to auras, lots of packets have changed, new features have been added. There is a whole lot of work to get something that really works for 3.0.2 and not only lets you log in. Yes, we could add support to log in with a 3.0.2 quite fast. But you would miss support for achievements, and all the other that came packaged with 3.0.2.