So, I had an interview with YoYo on Friday.
It was interesting, being the first interview I’ve been to in a number of years, and with the added bonus of anxiety… probably came across as a gibbering monkey! And a couple of questions put me on the spot and I froze – even though they were really rather easy. Oh well, they certainly made me feel welcome enough, and I’ll find out in a few weeks if I’ve managed to get away with my cheekiness of pointing out GM:Studio doesn’t run on Linux
Not being one to take nothing away from any interview, it made me rethink a bit on what I’m doing with GLESGAE.
I’m being rather lazy and relying a fair amount on STL.
Now, STL is nice and all, but it’s a specification that each compiler has to implement… and not all compilers are created equal. Earlier versions of certain compilers had rather spotty support for some STL standards, and being aware that just a simple call like std::string myString; can allocate more than you think!
Actually, std::string is a funny one if you’re not sure what it’s up to….
It does contain a bog-standard char* buffer for quickness. But, it’s rather small, and if your string is bigger than it’s buffer, it’ll allocate a new pointer *somewhere else* instead. Where that is, is anyone’s guess but the programmer shouldn’t really care… right? Of course, not all implementations will do this optimization, you may be unlucky and have it new some memory wherever it feels like anyway.
Dealing with consoles, you have a very tight amount of memory, and you need to know exactly what is allocating where and how much it is. It’s fairly common sense to keep track of your allocations anyway, but on memory constrained devices it’s paramount – especially when the bug bear of fragmentation rears it’s ugly head; not knowing what’s where and can be safely moved around is a nightmare waiting to happen.
Therefore, to prevent such nightmare scenarios, GLESGAE is being rewritten. Again. In straight C and providing my own containers for things.
Or more eloquently; there’s some nasty template hacking going on in JESSIE due to binding things with SpiderMonkey.. and it’s not nice. The only reason is for type safety, and while type safety is nice and all, having to perform odd looking things to make it work isn’t really worth it. It makes the code nasty to read, and a right bugger to maintain. GLESGAE has a few of these mad template things as well – and this is also what made the Resource System such an arse to deal with. So let’s just ignore all that and stick with C!
Of course, I still have other things to do in the mean time.
August didn’t go hugely well for the Badgers… we got our funding in, but due to schools just starting up and illnesses in the family, it’s been hard to get any work done. The Germies patch is a bit behind schedule, for instance.
It also means there isn’t anything coming in at the moment.. so to remedy that, September is going to be an interesting month for the Badgers. We’ll be working on three games – Germies, and two others – to try get something coming in.
We missed Ludum Dare, sadly.. which is a shame as we had an awesome idea for it, but hopefully we’ll get to the October Challenge instead and get something released then. And then by November, it’s back on to engine work in a mad push for a Christmas project.