At today’s meeting, we decided a testing architecture is necessary get to the bottom of our long-standing networking problems. Feanor had made such a testing program some time ago, so our plan is to update that to break the code down, layer by layer, to see how far down things are going wrong. This strikes me as a good step to take; we’ll get to the root of our problems instead of just patching holes when there’s still an underlying problem, without having to start from scratch or learn a new api.
If you’d like to help, we’re looking for people to work on this testing application as it goes through different stages. See the forums for details.
Before we work on the testing architecture we need to get the CSA build as bug-free as the older client ever was. Today feanor and I tracked down 2 kinds of bugs: the first were those due to incorrect math, which have been causing the ship to face in the wrong direction, teleport to a seemingly random location when you change the keys you’re pressing, and so on. It looks like these problems have been fixed, because the game works in demo mode again. The second kind of problem results in similar bugs, but is exclusive to games over the network. These are the next ones to fix.
A few days ago General Cade Smart noticed the ‘bot’ that was squatting on the domain www.dawnofinfinity.org had left. Geuis then registered the domain (again), so currently you can access this site through any of these urls:
http://www.dawnofinfinity.org/
http://www.dawnofinfinity.com/
http://dawnofinfinity.sourceforge.net/
We think the first one is most appropriate and easiest to remember. The actual content is stored on sourceforge, so if one ever goes down, or you want to save the most reliable bookmark, use the last one.
While we are still trying to figure out why ships are moving incorrectly, feanor has gotten the CSA client to build on Windows. Although not our top priority, syncing the Windows and Mac clients now has clear benefits. For one thing it means Windows coders (like Vulcan) can work with Mac coders on the game more easily.
Feanor is still hoping someone will step up to assist with the testing architecture. See his post on the discussion forum.
Confused? Don’t worry; we’re not. We had put some hacks in the engine to fix some bugs (having to do with the angle the ship points and flies). Now we (think) the bugs are squashed for good, but we haven’t had time to remove the hacks again. So since the hacks fix something that doesn’t need fixing anymore, they now cause problems. To use feanor’s words, it’s now “differently weird.” Once we take out those hacks the bugs should fly away to bug heaven for good. So far our strategy of methodically going through all the math in the code is paying off.
After a long meeting today, I’ve conceded that a re-write is worthwile. We’ve started a new module in CVS called uv, for unified version. We will build it from scratch one layer at a time in order to insure:
- low-level networking works before we add things like movement that rely on sound network code
- we use the same codebase for client and server
- the code is readable, editable, and executable by all, not just the person who wrote the code, and not just on some platforms
We’re also going to try using the Raknet networking library, which will hopefully handle what we need for DoI and save us from writing a lot of code. At least two coders are on spring break, so we hope to make some serious headway before we’re all tied up in real life again.
We’ve decided now is a good time to move our code. We’re moving in two ways. The first is to move from one sourceforge project to another. We use both “openevgalaxy” and “dawnofinfinity” over there, the former being a holdover from when we were called OpenEV. Moving our code to dawnofinfinity should remove the last remnant of our old name. This move won’t affect the website or forums, only the source code.
Our second move is from CVS to SVN. CVS and SVN are two different systems for managing code. CVS has been mostly adequate, but SVN is a popular new system that most of us see as an improvement. One of our developers has had trouble with CVS, so once we move he should find it easier to integrate code with the main branch. We’re trying to do both of these moves at the same time. Right now we’ve enabled SVN for both projects but are still figuring out how to actually transfer the code.
After everyone took the summer off, we’re once again working on the project’s code. The transfer to subversion, and dawnofinfinity instead of openevgalaxy, has encountered hurdles. It doesn’t look like we’ll be able to migrate all our code at once, but luckily we’ve decided to re-examine each file individually anyway. We’ve started working on the basic data structures we’re going to use, pulling older code from the csa and iserv modules. Using SubEthaEdit and AIM, we’re collaborating on one file at a time, adding it to a new collection that will eventually be a working new version of DoI. We’re trying to get this growing collection uploaded to sourceforge as soon as possible.
Along with our regulars, we’ve picked up a couple of new coders, like Rigdern, to help out. It’s always nice to have another pair of eyes looking things over, seeing if things make sense, so speak up on our forums if you think you can help us with the coding.
In other news, our wiki is down, because the maintainer, Geuis, is switching ISP’s. It should be back up shortly, with the same address.
There are several developments to report. As an aside, it’s only now that I’m updating the progress log, because I lost my account info for the site until Geuis helped me out. First, feanor and I successfully uploaded the code we’ve been working on to sourceforge. Unlike before, it was uploaded to the subversion server at the dawnofinfinity project, which should pave the way for putting all code there in the future. Secondly, cookies are working on our forums again. The problem turned out to be just a typo, which one of our newer coders Rigdern found. Thanks to him, you can stay logged in to the forums, see which posts are new, and all the other benefits of cookies. Lastly, we tried a new collaborative editor, called DocSynch, which runs as a plugin for the text editor Jedit. The problem with SubEthaEdit, which we currently use, is it isn’t crossplatform, so people on Windows or Linux can’t collaborate in real-time. Unfortunately we couldn’t get Jedit+DocSynch working with multiple users, so it might be back to the drawing board. But the thing to remember is now everyone can checkout the code on sourceforge now and work on the code by themselves.
It’s unknown when our old wiki will come back online. As a result, we’ve started up a new wiki. It’s a different kind of wiki (WikkaWiki) and is hosted at sourceforge, so it should be more reliable. The url is http://www.dawnofinfinity.org/Wikka/wikka.php?wakka=HomePage . The success of a wiki depends on its writers, so please help us fill it out. Thanks to Rigdern for picking a wiki that works on sourceforge and getting it up and running.
Of particular interest is a page for current tasks: http://www.dawnofinfinity.org/Wikka/wikka.php?wakka=CurrentTasks . This will be maintained as a short-term, focused list of what the development team is working on. If you can’t reach the rest of us and are wondering what comes next, this should give you enough information to get started.
DoI hasn’t seen progress over the last few weeks. Real life events have gotten in the way of me leading development. I intend to get things together again after Christmas. Until then, the developers are encouraged to continue on their own. We’ve put effort into the wiki, and I think at this point it’s capable of instructing a developer on what needs doing and how to do it.