12.12.2011

New ressource manager needed

I gave it up, the current ressource manager/tasker model simply won't work with mechanics that'll either require a fully customized tasker system (well, I designed the tasker to BE everything I'd need for such stuff!) or a shitload and not-to-debug synchronization objects. Somehow I have to control and bind these behaviours which's problem will only reveal themselfs when successlessly testing! I'll see whether I can use my beloved qmutex object for that. The problem isn't the parallel access but the fact there are gaps between the tasker logic (the only one which really works bulletproof in it's limitations) and the ressource manager logic, leaving room deviate the previously proper tasks when moving from a loading thread to a releasing thread. It sucks to learn this that late and I should listen to my concerns before attempting to implement something. That's what I really always re-learn these days: never simply start coding something that bears a significant, possible all-off-the-to-boats situation. And multithreading really is complex which 4 or 5 threads parallely running and possibly accessing task lists and randomly changing their priorities in a fixed access context. I feel a bit like I overdid it this time. But well, it's the first I'm really working with more than two threads in general and I'm starting to feel why there are dudes out there doing such stuff for their living. It's complex and open end. Something you either need to be experienced in or have a pro doing it for you.

Well, drawbacks always happen, but I don't want to stop here. The system's fully defined and I only need to make a brainstorming about a clever solution to eliminate the problems. Then I only need to define the loading and releasing functions and can start to port the rendered engine and try to get the physics working. I feel I won't be able to get everything done until February, the time were I have to start looking for an internship. *sigh* I was so eager to get something awesome done to show off my skills with. Bet it won't be more than what I already made of some years ago. I can't say I wasted my time inbetween, but I feel bad about having nothing to present, to show that I've actually done something. I actually created a lot of stuff, yes, but it's nothing I wanted to have to show off. I can't prove them good without proper use and presentation, so I'll probably have to work my way through annoying "side quests" until it'll get interesting. Or I'm once again overestimating the whole situation and they simply won't care for a having a game actually done? I mean I'll do my stuff there as a programmer, not a game designer. So my main concern is good tech and combining it, right? So we start by the previous problem of not having a completely assembled set of those...

I'm tired of this. Where did it put my synthesizer?

No comments: