[compacting]

D doesn't use a compacting garbage collector (and probably never will), so this is not really fair.

It says it does.

The GC works by:

   1. Looking for all the pointer ‘roots’ into GC allocated memory.
   2. Recursively scanning all allocated memory pointed to by roots looking for more pointers into GC allocated memory.
   3. Freeing all GC allocated memory that has no active pointers to it.
   4. Possibly compacting the remaining used memory by copying the allocated objects (called a copying collector).

Team Fortress 2, the definitive 3D art game.

Show me a benchmark where this makes a difference.

I wish I could, but I'm no GC developer so I've done no such thing. I'm just echoing what I've read around, specifically from the language D. It could be propaganda of course, but it sounds right.

That's how I'd do it, proper design in contrast. Preview:

Well, reference counting hidden behind smart pointers and stuff is a lot of overhead, and they can't handle circular references. Generational, copying garbage collectors handle all that fairly intelligently. There's also a lot of indirect performance gain from collection... for instance most destructors can be dropped entirely because they're not needed to free memory anymore, making the code faster overall. Collectors are also smart about when to collect - If you've got plenty of memory available, they don't need to do anything, whereas manual solutions don't have that level of sophistication. Then there's the issue with memory leaks; Sure you can argue they don't occur if you code properly, but collection guarantees it for no effort at all.

I also dislike garbage collectors for performance reasons.

Garbage-collected languages are actually faster than those not. It's counter-intuitive, but true. Modern generational copying collectors are quite efficient.

The black in the progress bars is easily identified as the background [...] the bars stand out more because they are a changing entity, while the background is static.

Sure it's easily identified, but that's the problem; You have to actually look at it and identify it. If it's unidentified stuff in your peripheral vision - as it is when you're reading thread titles - it draws way too much attention. Dare I say the game's name is more important than its completeness.

The backgrounds are not static. Some threads use these icons, others don't. Therefore the background changes from row to row - And that only reinforces my point. The background of other icons/no icon is white; Therefore the background here should be white as well, so that it does not change like you said.

Once again, contrast issue. The background cannot be dark because then less finished projects attract more attention. Make the background white.

http://www.desura.com/

I understand this is like an indie Steam.

I split a music track into smallish few second segments, defined a graph where each segment has probabilities to be followed by other segments (a probability-weighted L-system), then wrote a PHP script to output an M3U playlist for the lot. Result: Pre-rendered dynamic music.

Add real-time prediction of the appropriate next segment in a game and you have fully dynamic music. It's rather crude, just stitching Oggs one after another, but works surprisingly well given flexible enough segments of music.

I do not understand the point of URL shorteners. You copy-paste links anyway so it doesn't make typing easier. It doesn't save resources because a server is still required to do the short-to-long mapping of URLs. And worse, you can't tell where the link goes anymore; Could be a porn site, a rick roll, you just can't know. It also defeats the purpose of clean/SEO URLs.

Also, note that there's TortoiseGit for a nice Windows git GUI.

Specifically, these console devices have an interrupt called an hblank (the horizontal sibling of vblank) that is triggered when the display prepares to draw the next line. At this time, the game can shift the viewport to transpose the next line left or right.

Yeah I believe git doesn't diff (version track) non-text files at all. Furthermore it stores revisions in whole as opposed to incremental diffs (to my knowledge how SVN does it) so it shouldn't be effected at all - Except perhaps in that the repository will grow in size for each revised binary file.

Professor Mark Overmars made a massive mistake handing Game Maker over to these corporate clowns. I knew that even back then.

My serial key doesn't even work anymore. Not that I'd really use GM anymore anyway... but sometimes it might be convenient for a quick prototype.

[improve]

I voted no because you shouldn't make something different for the sake of making it different. If you can improve the interface then by all means, go ahead - If not, why fix what isn't broke?

I've yet to install all games/apps back since I reinstalled Windows so it's a little emptier at the moment than usual.

If I have to consult a walkthrough to progress, I consider that a failure on the game's part. Otherwise I stay far away from wikis (at least until I've completed the game once) - I've had the pleasure of accidentally spoiling massive plot elements browsing a game's wiki before.

Revenge of the Titans