Is This True?

[unmerged(151058)]

I suppose there was a need for a proper scripting language (those cute paradox style txt files only go so far ) and lua was chosen, probably because it can be integrated easily into other projects (the game, that is). Take a look at wikipedia for an overview, I haven't used it before myself.

 

[quetzilla]

AI uses LUA

Correction: SOME aspects of the AI use LUA. It's mainly for trading and foreign policy, as well as handling the tech and intelligence screens for the AI. The majority of the AI is not handled with LUA.

 

[Kriegsspieler]

It is a scripting language. I can't tell you why Paradox decided to use it, ask them. And .lua files are lua script files. Kinda self-explanatory don't you think?

Indeed it would be, if one was in a position to know what lua was in the first place!
I did look it up, anyway, and Luka's reply put it in context too.

 

[telesien]

This kind of reminds me the time, when I asked if the goal oriented AI is now (or in any reasonable time in future) able to work properly

 

[unmerged(154327)]

And there you go! I love playing 1.3 personally, it's great fun compared to 1.2.

What exactly do you love about the game vs. 1.2? I admit, I like to experiment with the units, but operationally and strategically the game is still broken (among other issues.) Other than being much faster than 1.2 I am just curious what fun are getting out of the game so far?

 

[Scryder]

What exactly do you love about the game vs. 1.2? I admit, I like to experiment with the units, but operationally and strategically the game is still broken (among other issues.) Other than being much faster than 1.2 I am just curious what fun are getting out of the game so far?

Well OK, here goes. First, I haven't played any other HoI game before, so I don't know what the "ideal" game is. I'm currently liberating Europe as USA, with full manual control. In 1.2 I managed to defeat Germany without any sign of resistance in less than 2 months. In 1.3 as soon as my troops landed, I faced much stiffer opposition and I found it a struggle advancing every step of the way. I had to pull my defensive lines back a number of times and regroup before advancing forward. A couple of times I got ahead of myself and got a couple of Mechanized divs trapped behind enemy lines. I am aware that the AI is not perfect, but their troops did manage to cut off these divs quite well.

My troops have yet to run into supply problems either, even when they were fighting in the low infra parts of Italy.

I know there are still bugs, I report them in bugreports if I can reproduce them so they can be fixed in a later patch. Overall I'm getting a much better gaming experience with 1.3. I don't know, maybe I'm just easily pleased...

 

[Neumann]

Complete bollocks from some random idiot on the internet.

"there's no way it can crunch the numbers in a world war II game that has it doing 10,000 provinces of weather, troops, terrain, units"

LUA is not used for any of those.

Excellent POST! Well put!

 

[dublish]

You must be drunk, mad, crazy or correct.

Why must the four be mutually exclusive?

 

[womble]

Complete bollocks from some random idiot on the internet.

"there's no way it can crunch the numbers in a world war II game that has it doing 10,000 provinces of weather, troops, terrain, units"

LUA is not used for any of those.

Thought not. LUA's used just as the presentation layer, neh?

 

[mbb]

Is this true? What can we do to fix this?

The post was really talking about the performance of the game. There may be some truth in there.

Scripted languages tend to be slower than compiled languages, often are designed with a specific purpose in mind and not good at performing certain tasks depending on the task and language. Whether this is true in this context or not - well, Johan has had his say.

There is also the added complication that the LUA script is interacting with the executable and it is impossible to judge the use of the LUA scripts without having a very good knowledge of the game engine itself.

I was surprised to learn that LUA was going to be used for parts of the AI.

I understand the requirement to make the AI moddable, but a scripting language was the last thing I expected to see for a part of the game that I think needs as much performance as it can get.

Are there any benchmarks for comparison?

But what was the true intention for making the AI moddable in this way? Was it for the modding community or for the developers (internal/external) themselves? Because if it were for the latter's benefit, then what was wrong with C++?

If there are certain LUA scripts that continue to be a problem performance wise then there is always the option of porting that functionality into C++; but that would be Paradox's call and one less thing for the modders to play with.

The community can point out inefficient LUA scripts - I'm sure that has been done already. But we can't change the nature of the game's construction. That's up to Paradox and I suspect it is pretty much set in stone by now.

But to say the game cannot be 'fixed' is just not true.

 

[podcat]

It is a scripting language. I can't tell you why Paradox decided to use it, ask them. And .lua files are lua script files. Kinda self-explanatory don't you think?

http://en.wikipedia.org/wiki/Lua_(programming_language)

Pretty much the standard reasons you pick a scripting language. Letting modders have more control and do things you didn't anticipate in advance and speeding up development of certain stuff that requires quick iterations. Most games use some for of scripting language for AI or other things (Civ4 uses python heavily for example).

Lua is popular because it is pretty minimal feature wise, simple to learn for people with little programming experience and pretty fast for a scripting language (lua is about 20 times slower than compiled C++ code for pure calculations), but we still put most heavy calculations outside lua code (in C++) and export them for easy calling from lua to speed things up and still keep the moddability.

oh yeah and as johan said the quote in the first post is just your regular mindless forum drivel. When someone claims to know "secret" tricks to doing something and doesnt explain, or even hint at it, its usually just bollocks.

 

[Kriegsspieler]

Lua is popular because it is pretty minimal feature wise, simple to learn for people with little programming experience and pretty fast for a scripting language (lua is about 20 times slower than compiled C++ code for pure calculations), but we still put most heavy calculations outside lua code (in C++) and export them for easy calling from lua to speed things up and still keep the moddability.

Podcat, what I don't yet understand is how the adoption of Lua has changed the realtionship between "text" files and the game engine. In all earlier P'dox games, you had .txt files that provided data to the parts of the engine where the calculating work was done. Here, though, it seems that the relationship has changed, with more of the actual crunching being done in moddable portions of the game. Is that right?

 

[Chaingun]

Podcat, what I don't yet understand is how the adoption of Lua has changed the realtionship between "text" files and the game engine. In all earlier P'dox games, you had .txt files that provided data to the parts of the engine where the calculating work was done. Here, though, it seems that the relationship has changed, with more of the actual crunching being done in moddable portions of the game. Is that right?

Yes, pretty much. When done right it should still only be an insignificant part of the execution time.

There's an old saying that 10% of the code accounts for 90% of the execution time. Most time will be spent inside inner loops, and if those loops are implemented in C++ and exported as functions to Lua, performance will not be hurt by much.

I think there was great pressure to get this game to market and LUA was choosen because of the ease to implement it quickly. It wasn't a practical choice. LUA has never been used in a ANY game for beyond simple configuration (WOW) simple pathfinding (Shooters) and simple NPC control (Baldur's gate- which BTW had horrible results

My favourite counter example here is Dawn of War 1, where the entire AI was written in Lua. As a result, modders could improve it to almost play as well as an experienced human. (The original unmodded AI was pretty lame.)

Also weird he mentions pathfinding because that's something you'd not put in Lua if you want performance.

Aside from that, the rest of his text is bull.

 

[unmerged(49695)]

So, er, what DOES LUA do?

means Moon on portuguese

 

[sbr]

Why must the four be mutually exclusive?

Speaking from experience? :rofl:

 

[podcat]

means Moon on portuguese

yeah it was developed by some brazilians.

Podcat, what I don't yet understand is how the adoption of Lua has changed the realtionship between "text" files and the game engine. In all earlier P'dox games, you had .txt files that provided data to the parts of the engine where the calculating work was done. Here, though, it seems that the relationship has changed, with more of the actual crunching being done in moddable portions of the game. Is that right?

pretty much yeah. The history files with triggers we used for most stuff, and still do is pretty much a declarative scripting language itself, so the difference is really just syntax and how powerful/flexible the language is.

 

[unmerged(148805)]

For the record, Eagle Dynamics uses LUA to great effect in it's flight sims Lock On, it's expansion pack and a follow up product concentrating on the Ka-50 attack helicopter (Black Shark).

The latter is THE most complex and realistic flight sim I have ever come across with regard to flight model and dynamics. LUA causes no performance issues there and I seriously doubt that its responsible for any performance issues in HOI3 either.

 

[unmerged(40080)]

So . . . what of Johan's categorical remark about it not being used? t