[IPAndrews]

20. Random number SEXPs that make use of a random number seed that the user can set. Such that if a mission is called twice, and each time the same random number seed is used, any calls to random number SEXPS (or send random message) give the same result. This would be enormously useful for pseudo randomly generating missions.

[Goober5000]

20. Random number SEXPs that make use of a random number seed that the user can set. Such that if a mission is called twice, and each time the same random number seed is used, any calls to random number SEXPS (or send random message) give the same result. This would be enormously useful for pseudo randomly generating missions.

What about setting a variable equal to the rand sexp when the mission starts, and using that variable for each "random" number requirement?

[IPAndrews]

Use a load of random number variables to store "rand" values? Well okay:

1. The random number seed solution is more elegant and uses less variables and sexp tree nodes (both a limited resource).
2. rand generates different numbers each reload of the mission. Not the functionality I'm requesting.
3. Generate the numbers in the previous mission and pass them? I figured you'd think of this like I did. Works too well since the rands in the previous mission seem to generate the same numbers each time. Replay previous mission also scuppers this idea.

Bottom line, the seeded random number functionality is there in most languages and it's incredibly useful. You can build whole galaxies with the principle. Check out Frontier (Elite 2) or Damacles. At this point it would also be the single biggest improvement that could be made to help my BHX campaign. Okay maybe after the modest sexp tree node limit increase (or dynamic memory allocation for sexp tree nodes).

I realise I'm a pain in the ass but try to humour me. My crazy ass campaign is coming on leaps and bounds and with a little co-operation from the SCP team will soon be done. Once it is, it'll pave the way for others to rip out my "code" and build there own pseudo randomly generated stuff. It's cool. Trust me  .

Edit: Doesn't FS already use seeded random numbers? I'm guessing the seed is set based on run-time or something. That's usually how it works. So surely this request just involves the implementation of one set-random-seed SEXP?

[Goober5000]

Edit: Doesn't FS already use seeded random numbers? I'm guessing the seed is set based on run-time or something. That's usually how it works. So surely this request just involves the implementation of one set-random-seed SEXP?

Actually, I'm pretty sure it's not a random number at all, but rather something calculated from mission time.  Try displaying the results of rand in conjunction with has-time-elapsed and see what you get.

[karajorma]

1. String -> number conversion sexp.

Got bored. Added this. It was only about 2 lines of code

Little less certain how easy it's opposite will be to add without actually having a proper string return type. Will investigate when it's not 1:30am

[WMCoolmon]

Proper string return types would be handy for script-eval-string.

[IPAndrews]

Got bored. Added this. It was only about 2 lines of code

What? Really? One of my suggestions made it in? This is a cruel joke right? 

[IPAndrews]

Actually, I'm pretty sure it's not a random number at all, but rather something calculated from mission time.  Try displaying the results of rand in conjunction with has-time-elapsed and see what you get.

Well no random numbers are really random numbers. We all know that  . This is just another variation on a theme. A fairly odd one though. Is there some reason it works like this? Some random event somewhere in the standard campaign that is always run at exactly the same time and requires the same random number to be generated? If so why the heck is a random number being used? The more I find out from you guys about the workings of this code the more baffled I get.

Well to be on the safe side - I guess - new rand-seeded and rand-multiple-seeded sexps would be required. If it was me, I'd just modify the default behaviour of the existing sexps because I would bet good money (albeit someone elses money) that behaviour is just wierd programming rather than serving any purpose. A random number seed set on the milliseconds of current time at mission start would be sufficient, and allow it to be modified in order to do cool stuff. 

See what I did there? I nodded to subliminally influence you to agree with me. 

[karajorma]

What? Really? One of my suggestions made it in? This is a cruel joke right? 

Nope. It's in CVS already. All it basically does is call the standard library's function for doing the same thing.

Incidentally Goober seems to have snuck in a Key-reset-multiple SEXP when I wasn't looking

[IPAndrews]

Incidentally Goober seems to have snuck in a Key-reset-multiple SEXP when I wasn't looking

So the original key-reset only worked the once? I can't say I ever noticed but it would explain a few things.

[karajorma]

The original key-reset displayed all kinds of odd behaviour. I had a thread on it here

EDIT : Added the pics again

[IPAndrews]

I like the subject title. Looks like I'm not the only one who gets stressed with this stuff.

[karajorma]

Told you I recognised the symptoms

[StratComm]

Well no random numbers are really random numbers. We all know that  . This is just another variation on a theme. A fairly odd one though. Is there some reason it works like this? Some random event somewhere in the standard campaign that is always run at exactly the same time and requires the same random number to be generated? If so why the heck is a random number being used? The more I find out from you guys about the workings of this code the more baffled I get.

Truth be told though, there's no such thing as a truely random number generated by a computer.  You can do things like seed the generator based on the system clock or something like that, but ultimately it's still not random.  In this particular case, it's probably just setting the seed improperly rather than doing anything wrong.

That said, since Rand has been non-random for as long as it has, it becomes a question as to whether it's better to fix the SEXP that's there and risk breaking a campaign that's built with it or to create a new actual-rand sexp that behaves more like you want and doesn't affect backwards-compatability.

[IPAndrews]

That said, since Rand has been non-random for as long as it has, it becomes a question as to whether it's better to fix the SEXP that's there and risk breaking a campaign that's built with it or to create a new actual-rand sexp that behaves more like you want and doesn't affect backwards-compatability.

It is possible to have the best of both worlds here. I cannot think of any way that changing the original sexps to use a seed generated from current time at the start of the mission would affect backward compatability. After all, you're not changing what the sexp is providing, just it's internal workings. Once changed all that's needed is a set-random-seed sexp to override the one generated at start of mission and suddenly you have random numbers which are more random  and less predictable in general but can generate numbers is a predictable order by overriding the seed  .

[karajorma]

Actually, I'm pretty sure it's not a random number at all, but rather something calculated from mission time.  Try displaying the results of rand in conjunction with has-time-elapsed and see what you get.

Nope. It does generate a random number. It calls the rand_internal function which simply uses the standard C library's rand() function.

 

20. Random number SEXPs that make use of a random number seed that the user can set. Such that if a mission is called twice, and each time the same random number seed is used, any calls to random number SEXPS (or send random message) give the same result. This would be enormously useful for pseudo randomly generating missions.

Done and in CVS.

I haven't looked at random-of yet but Rand and Rand-Multiple now take an optional 3rd argument which allows you to seed the functions. I don't think I've changed anything if the seed isn't present (it should do exactly what it did before if no seed is present).

[IPAndrews]

That's fantastic news. You have no idea how useful that will be! Well to be fair, you probably do actually. No more reloading a mission to generate an easier one. This is truly great news. Must go tell Madaboutgames now .