Mind Forever Burning

Which was exactly my point. I think the reason C++ lacks them is that it's harder to distribute and integrate well-crafted services than in Java because of design decisions around memory management.

Anyway, I'm letting the thread get back to its original, intended tangent at this point. :P

I disagree. C *only* has malloc/calloc/realloc/free; those are the standards. C++ has new and delete, but also many different varieties of wrappers on top of them (and of course malloc/free themselves). In many cases, when integrating frameworks, you have to disentangle the thing that was built on top of new/delete and then rewrap it for the benefit of the next layer in the stack. C++ operator overloading is a bit of a double edged sword here because it's possible and very tempting to make your own syntactic sugar that is almost, but not quite, entirely unlike the one your neighbor made.

The STL containers are neat, but asking everyone to be as smart as Stroustrup in order to make rational re-usable classes is unrealistic to say the least. One of Java's biggest successes is that it assumes the average implementor is much, much more naive than the language designers.**

Automatic Reference Counting. It's a language feature built on top of notational conventions in Objective-C that allow the toolset to automatically generate the appropriate retain/release semantics, so the developer doesn't have to do any explicit memory management but you still get deterministic object lifetime semantics. It advertises being both headache-free and GC-pause-free at the same time, which is a huge win for mobile development.

** Which bothers to no end those implementors who are actually smarter than the language designers.

Try replacing the period at the end of Arcanaville's sentence with a question mark. Or, here:

"Can X even move the needle much at all?" where X is "what you do in any number of hours on a weekend"

On the subject of C++, I love the language; it's still the only OOP implementation I know where I can "feel" how the code I write will be represented on the machine. (Java is close, but the lack of determinism in object life cycles hurts.)

But C++ has a huge Achilles' heel when it comes to large projects, and that is the lack of a standard mechanism for managing heap objects. Smart pointer? std::auto_ptr? Manual new/delete with clear ownership? Something custom? While it's obviously a solvable problem locally, when you start to integrate large systems with different ideas of "standard" you get messy interfaces, difficult bugs, and general lack of developer efficiency. Because of this, frameworks become difficult both to write and to integrate, and you end up with C++ seeming very support-poor when it comes to modern application development.

I want to learn more about ARC because it seems like something of that nature could be applied to C++ also, and if it were adopted broadly and incorporated into all the tooling, it could really go a long way toward solving this issue. But I don't know enough about ARC yet to say for sure, and of course filtering something like that through the C++ specs would take considerably longer than Apple just declaring it by fiat.

Sorry for the thread-jack, BillZ.

I used to dislike Obj-C for its syntactical confusion between function calls and method calls, but once I got over that, I started to see quite a bit of inner beauty to the language architecture. It seems to neatly sidestep a lot of Java's heaviness in reflection, for example.

I'm not aware of any projects running Objective-C server side though. Is there a language implementation that would run on Linux? Xserve got the axe a few years ago, so I'm not sure what hardware you would run. (I once asked what Apple uses internally to run automated tests that require server side OS X, and was told that they use HP racks running custom, non-public OS builds.)

Actually, I wasn't quite sure whether you were referring to cases in the data where values were interpreted differently depending on context, i.e. "fooBar:1" when you are fighting in Grandville means you do a backflip animation while "fooBar:1" when on the last map of the first villain respec trial means you do a double-twirl and end by planting a kiss on your current target's forehead.

I don't know if such cases exist (I no longer have much more than passing knowledge of how the game's internal mechanics work) but I've definitely seen real world cases of it. Mostly they seem to happen because people don't want to change a database schema when they add new functionality, so they take parts of the parameter-space that are currently unused and make them mean something different.

While I agree with you, I have a great deal of personal sympathy for the devs on this one. Having now tried to implement a game of my own, I'm very aware of my own limitations, and giving players direct access to (potentially quite flawed) mechanics can feel a little like letting the mike boom drop into the camera frame in a movie.

Put another way: if you look at CoH as a strategy game, then failing to reveal mechanics seems like a huge blunder, but if you look at it as an interactive story-telling experience, it starts to make more sense.

Back in 2000, I was tasked with implementing security for a micro feature within a larger software application. I put in a basic authentication handshaking protocol, and stubbed out the encryption part of it with a very dumb, very obviously reverse-engineerable algorithm. I won't even repeat it here, it was so ridiculous.

Five years later, the stub had still never been replaced with real encryption the way I had intended. That year users found a way to trivially gain access to the micro feature... not by reverse engineering my ridiculous toy algorithm, but by simply swapping versioned components from other installations, defeating the higher level logic that decided whether the encryption handshaking needed to happen at all.

There's probably more than one lesson in there, somewhere.

I realized earlier today (and confirmed a few minutes ago) that the wrapper I made for the binary fails if the path has spaces in it. Rookie mistake! Anyway, a fix is forthcoming.

Also, due to the way I'm requesting the permissions required to run the program, the output files are written with root ownership. I might be able to find another way to do that, so if this becomes an issue for people, let me know.

I think it's reasonable to imagine that it's not the same decision makers. The devs still control the game code.

Of course, they might no longer be able to push an update.

I had heard as much, but I just couldn't believe NCSoft would do such a thing. I managed to re-up for VIP status on Friday; I guess I slipped in under the wire.

That is just... I have no words.

It's been so long that I had forgotten that I did most of my posting on this forum under a different name, Protea. Then for a while I was KeepDistance. I changed to Mind Forever Burning when the Mission Architect came out because I thought it was a better pseudonym to write under, but I stopped playing the game not long after that, so... yeah. Some of you were probably wondering who the heck this person was who was getting all misty eyed over a forum they'd never posted in.

I'd save the praise for when I actually produce something. I got an initial inkling of how they're doing this and it's not trivial.

*raises hand*

Send me the source code and I can take a crack at it. PM me for info.

Can't you just fight in Warburg?

Sometimes I have to get a bad story out of my head before I can get to work on a good one...