Game Design, Programming and running a one-man games business…

Today’s worklog

8.00AM: Clement Freud has died. He gave me an award at my secondary school many years ago. He was very short. Also he was very funny

9.10AM: All the business email stuff is dealt with, time to start work on putting one of the new ships into the game, one I have declared to be the ‘Imperial Centurion Cruiser’. It looks badass

9.50AM: Most of the graphical stuff is done and dusted. Now I just need to add in the data defining this ships damage textures and module slots, and get it tested and in game

10.30AM: The Centurion cruiser is done and looks great in game. Took a screenshot to give feedback to the artist. Everything seems to look ok, so I don’t need to go back over any artwork. Plus the multiple beam lasers look fantastic. I’m pleased

11.10AM: There is definitely a shutdown problem with the game. Dev Studio churns for ages. Must be a huge and new memory leak somewhere. Damn.

11.50AM: It’s definitely a bug with the main battle screen. But where? The sound? graphics? AI?

12.00AM: Interrupted by door to door salesman, thus killing my efficiency. Time to release the hounds.

12.20PM: I suspect this isn’t a memory leak, but just my debug output backing up bigtime because every time I create a blast glare effect (also used for non-penetrating shield impacts), I tell the debugger about the texture I used. Clearly this code is inefficient anyway, because with dozens of the smaller glares I need to have that texture cached, not be looked up so often. The game now shuts down fine in debug, but I need to fix the underlying inefficiency. In fact, it’s worse than I thought. I need a newer, efficient system for shield impacts

12.35PM: Works fine but forgot the original texture was a quadrant and this one needs not to be. redoing it… Also just spotted a graphics glitch with the conĀ  trails pulsing on and off.

12.45PM: Having crash issues when trying to do some profiling, Grrrr. I will mull it over some wii-jogging and a cup of tea.

1.30PM: Still can’t get profiler to not crash. Time for tea.

2.20PM: it now crashes somewhere else. progress?

2.50PM: Sod it, I have a workaround, and bugs to fix. some lights in the wrong places, and an idea on how to make drawing running lights faster if they are zoomed out. (check rendered size > 1 pixel BEFORE checking if onscreen, which takes longer).

3.30PM: The debug copy no longer links. some obscure ANNOYING linker error to do with debug and not debug Microsoft crap. Very tedious and I’m not getting actual game stuff done :(

4.14PM: Think it’s fixed, plus I tracked down the flashing con trials bug. It was me not setting the render mode correctly before I drew them, and them getting drawn early if they filled up a vertex buffer before I was ready to render them all. Good old Nvdia perfHUD was my saviour. Now back to coding fun stuff.

6.25PM: Finally got the linking stuff fixed by a total wipe and rebuild. Grrrr. Next on the agenda is adding some better parallax effects, which are a nightmare with 2D + zoom + variable screen res and scrolling.


6 thoughts on Today’s worklog

  1. Really interesting, I’ve been following the development and really interested by the process. Thanks for sharing!

    I’m currently teaching myself C++ from a book… Not really much talk of errors.

    It sounds like it’s a regular occurrence to get problems with the compiler and linker with subtle bugs. Seems quite frustrating. How does one learn how to fix bugs, errors and memory leaks, I’ve come across some really obscure linker errors while working with really small learning projects with only several classes so bit worried about how I’ll work out how to fix worse errors as my projects get more complex.

    Also wondering, how long does it take for you to do a normal build? How long for a wipe and rebuild?

    Cheers, and can’t wait to play!

  2. A full build is maybe 10 minutes, so it’s ok as long as you aren’t having to do it ten times in a row to track down exactly which compiler setting you need, which happened a few times today :(

  3. to Ranoka:
    There are a few issues with compiling and linking, but you’ll eventually get over that in some time. Everyone went through that. It only takes practice.

    to cliffski:
    10 minutes? That’s too much. Maybe your source code consists of too many small files?
    If you have dual-core CPU, use /MP multiprocessing switch, it will halve your compiling time. Or even faster are ‘precompiled headers’.
    Btw, did you ever think about releasing the source code? :) (it would get you some media attention too)

  4. Man….

    A) You code really fast !!!

    B) Just shows how much work goes into these “little” games !!!!!! No wonder you can only so few a year. Ever thought about off loading some of the grunt work to India / China / Israel so they can work over night because of the different time zones and then have some stuff back to you ready in the morning?

    C) Would it be too hard / expensive / mind blowing to develop the game on two PC’s? so one could complie why you code on the other ?

    I really need to get back into coding – Sooo out of date !!!

  5. Btw take a look at Visual Leak Detector. It will tell you the exact line where your memory is bleeding.
    It takes about 40% of performance for me, and %% grow very fast, when you have lots of new/delete – which I consider memory-fragmenting bad coding habit ;). And of course make sure you pass function arguments as const references to avoid useless copying or class constructors/destructors.

    Let me know, if you already know all this. :)

Comments are currently closed.