(Review Completed) [SP] Un-Impossible Mission

[macguba]

Ah, it's more than possible.   I thought all the groups were minimum 3, but maybe not.   It could be a mistake, but it could be a spot of cunning editing from the past so I don't intend to change it.  *checks*  Ah yes, that one was at one time accounted more difficult than some others so I reduced the probs there.

In that case your analysis will give slightly lower numbers than mine further above.    My definite loon has a prob of about 55%, so my overall figures will tend to be about 1.5 loons higher overall.

I know nothing about C, sadly.        

[THobson]

You don't need to know C - the program is written - you just need a compiler.  There are some free ones about, I just can't lay my hands on one at the moment.

Can anyone please give mac a C Complier?

[THobson]

I know nothing about C, sadly

No need to be sad.  It is a bloody awful language, but the best I have got at the moment.

I don't suppose you have Microsoft Visual Basic?  You would need either the learning edition of the developers edition (I only have the learning edition so I am not able to distribute programs)  I would convert the program to VB if you could run it.

[macguba]

I used to know a little BASIC.    And I once wrote a couple of macros in Excel.   Other than that my programming skills are strictly confined to OFP code.

Would be nice to be able to do the analysis though.   In particular to play with it a little.    The 90-70-70 came from a massive spreadsheet (subsequently ceremonially burnt) testing everything in multiples of 5.    

The other news is that the Abrams is staying in.   I've used a couple of switch triggers to swap it between guard and S&D waypoints:   the guard sends it to you and when it arrives and spots you it goes to S&D.    Back to Guard if he loses you.   (The Bradley group have been given the same routine.)

I've already fixed it so that his accompanying Bradley is always present, and the dismounts in the Bradley actually get out.   The version you're playing has them trapped in the back.   ;D

The extra triggers have been offset by removing 12 ammo crates and some sandbags.    init.sqs has a bundle of addweaponcargo commands so for the first time you really do have total ammo resupply available.    It turned out, of course, that the HK is the only primary weapon that doesn't have its mags in a default ammo crate.

The next version is basically ready - I just need to check it still works.   This mission is so damn complicated now, I really worry about knockon effects of changes.

[Sneaker]

Well ive just read the last few posts as if it was in swahili, but anyway i think the consenses was about randomness and the whole train of thought was brought about by the start of the mission.
Sorry to backtrack a little , but here goes:

First to the point of tanks,
You could have a script to detect when theres no support around it and when there isnt it doesnt stay around too long, or as youve said combing a search and destroy WP.
As for for crew destroying tanks im not so sure about. You could do it, although i would be more against it. If it was done, id rather have some warning, for example create an A10 LGB thats called in to stop the equip falling into enemy hands by bombing the tank(s) and area, so youve at least got some warning and a chance to exfil.


Now for the other stuff,
I dont agree with everything that is said about random and I personally dont think it should be easier at the start.

If its possible to only get 1 squad in your location, then even if you get everyone appear and reinforments come in fast at the start, it might be opposite at different areas and not so many loons about, therefore you cant complain about randomness because its a balance, even if its possible to get unbalanced

On principle, 99 times out of 100 i play until i die. In other words i dont restart until i bite the bullet. This is mainly because of it can be too easy to get into the frame of mind if someone dies or something even goes slightly wrong, you go for the restart button. I prefer to play on cause shit happens, too bad. This meant i had a couple guys die at the start, which would technically make it harder. However because of the point i made above, it doesnt make any difference.
Theyre gonna come either way, so its not random behaviour, (mainly due to constraints with the AI, but also waypoints)

Although sometimes frustrating, most of the time later on in the mission was of suprise, such as crawling right past a squad and not realising it was there until you are on your way back from placeing some satchels. Without so many variables on things like positioning this wouldnt happen, and the mission would be worse off.
By adjusting small parts and trying to make them more friendly ,even just  the start, would affect the mission as a whole, i think negativly.

 - if the 'randomness' follows rules, its not random!

[macguba]

Well ive just read the last few posts as if it was in swahili, but anyway i think the consenses was about randomness and the whole train of thought was brought about by the start of the mission.

;D  Your Swahili is better than you thought. :thumbsup:

First to the point of tanks,

The tank thing is settled for the time being.   I don't feel comfortable in this mission having crews destroy tanks.   I think its a fantastic idea for a general mission, but here I don't want anybody to say "I could have done it apart from that special script".   (Even if it's bollocks)  The solution I've worked out - just waypoints,  triggers, setWPPos commands and variables - is within the design parameters and seems to make a difference.    The armour does tend to sit around a bit while on S&D, but once they lose sight of you they go away instead of sitting around waiting to be boarded.

I dont agree with everything that is said about random and I personally dont think it should be easier at the start.

An argument, cool.       Sorry discussion.  

it might be opposite at different areas and not so many loons about, therefore you cant complain about randomness because its a balance,

You are correct.  And this mission does have a habit of doing you a favour and then making you pay for it.   I suspect THobson was just unlucky when he did the start.   The random genertor in OFP is not that good:  when you get one result you tend to get it again on the next throw.    Thus, if you have a hard start and retry, you'll probably get a hard one again.    This is most obvious is start locations - if you hit retry again and again and again you will restart at the same place as the previous try far more often than one time in eight.

On principle, 99 times out of 100 i play until i die. In other words i dont restart until i bite the bullet.

This is good from my point of view - beta testers with different playing styles are vital.

meant i had a couple guys die at the start, which would technically make it harder.

Obviously it is slightly harder if you have fewer loons, but actually it doesn't make that much difference.    The mission is so hard that as long as you have a three or four loons to watch your back, carry ammo for you and fetch vehicles that's all you really need.      You've got around 350 enemy loons to drop to clear the island:  even if your guys get 10 or 15 each before they die it they can't win the mission for you.

By adjusting small parts and trying to make them more friendly ,even just  the start, would affect the mission as a whole, i think negativly.

The start groups are not just the start groups, they form the bulk of the opposition.    They are the endless waves of groups that just keep on coming and coming.    Most of the other groups have specific jobs, such as snipers, armour, road patrols, etc..    Small changes in these start groups do not cause any significant difference to the mission as a whole, although small changes can affect the start significantly.   Large changes in these groups obviously do affect the mission as a whole.

Having looked at all these numbers at present I am minded to leave the start groups as they are.   Originally I was very concerned by some of THobson's comments, but having done the sums I'm a bit easier in my mind.     I've lost track of exactly what was changed when, but I did make the start slightly more difficult because I wasn't getting enough complaints about it:  in the great spurge of beta testing the mission enjoyed a few weeks ago everybody made it past the start without too much trouble.     I think now its probably at about the right level overall, though I'd like to smoothe things out a little.    The dips in the graph are not attractive, and it's too steep around 10 and 20.

[Homefry31464]

You don't need to know C - the program is written - you just need a compiler.  There are some free ones about, I just can't lay my hands on one at the moment.

Can anyone please give mac a C Complier?

If you send me the code, I can most likely get it compiled... we are talking C++ here, correct?

[THobson]

Homefry:  
Thanks.  I can compile it.  The reason mac would need a complier is so he can change the probabilities and the locations etc. for his current (and possibly future) version.  No we are talking C not C++.  

macguba:
I will have a look at some of the other starting locations tonight.  I have a suspicion that there may be some differences.  

I've already fixed it so that his accompanying Bradley is always present, and the dismounts in the Bradley actually get out.  The version you're playing has them trapped in the back

You are now telling me that the version I am playing is not difficult enough!!

However, the key question is how to smoothe it out a bit.    The ideal is of course a more or less normal distibution.

Well err...

if you introduce some dependencies you might be able to do it.

[macguba]

You are now telling me that the version I am playing is not difficult enough!!

Yep.     (Only marginally though.    You had the Bradley anyway (it was on 75%) and there were infantry around at the time ... another 3 or 4 wouldn't have made much difference and you might well have popped the Brad when they were still in the back anyway.)   This is a feature of the beta testing:  each version tends to be harder than the last because bugs get fixed.   I do introduce compensating changes from time to time, but of course every time somebody completes it I am tempted to hike the all the probabilities a bit.

Actually the key question is, at this stage of the proceedings can I face making the changes?    Probably not is the answer.   I can't use dependencies for the start groups at this stage anyway - that would be a huge change in the mission which is something I don't want to do at this stage.  I'm just desperate to get the wretched thing finished.      For example it took many hours to figure out where each group goes after the start and introducing new, conditional groups ... well I can't face that again.    I've published two final versions already, and the next one is not final as it is.

What I might be able to do is adjust the group probabilities, or adjust the individual loons probabilities, to straighten things out a little.

[THobson]

well I can't face that again

I can understand that.

What I might be able to do is adjust the group probabilities, or adjust the individual loons probabilities, to straighten things out a little.

I don't know if I would do that.  You have three peaks in your histogram - so what, the mission works and you would not have known if I hadn't done the simulation, I would not worry about it.  Rather than make more changes you might want to just want to learn to live with the distributions you now have.  

I was conscious though that you were looking at the randomness with consistency as a more generic issue, not just related to this mission.  In that context building in deterministic dependancies rather than allowing total randomness I am now (as a result of the last 24 hours) convinced is the way to go.

[THobson]

Sneaker

i play until i die. In other words i dont restart until i bite the bullet. This is mainly because of it can be too easy to get into the frame of mind if someone dies or something even goes slightly wrong, you go for the restart button

I can see that and I agree with it, if I could also offer an additional perspective: if I get to a point where for some reason the mission is now un-doable (at least by me) it feels a bit like cheating to continue and so discover more about the mission design and then take that knowledge back to an earlier part of the mission.  

I guess the point is, however we play we need to be honest with ourselves.

I was impressed by how you completed the mission.  I am much more dependant on having a healthy squad around me.  Mac made a comment about how relatively few loons the squad will take out.  With my way of playing I spend much more time spotting than I do shooting.  My squad does most of the work.  I put my hand/eye coordination issues down to age - that is my excuse anyway

[Sneaker]

Mac

Obviously it is slightly harder if you have fewer loons, but actually it doesn't make that much difference

That was , infact my point  ;D. I said something about it not making any difference in the following sentence although it mabye wasnt clear what i meant lol - it doesnt make a difference beacause like the variables, its just a balance - few may die later, or you might getyour whole squad through at the start, then run into an mbsuh two minutes later anyway

Thobson

I am much more dependant on having a healthy squad around me

So am i . Indeed it would have made it easier if i did, and i used them i thought reasonably well just to do the first objective, positioning them to see where i couldnt etc.. the difference being most of them were dead by the end of it lol,
When i saved at the various points, i definately wasnt wanting to start from scratch, and also at that point i didnt know how many loons were left lol.
Later on , without civvie, id problaby have taken longer, or still be trying to complete it - hes da man!
In Hindsight, i have thought i could have used one less save, since i used two on the hillside, but because of the variablity of the map, chances are that if i didnt id run into some patrol at the most inopportune moment.

it feels a bit like cheating to continue

Indeed i was only commenting on different playing styles and theres no point on conituing if you dont want to, and i do the same. With both styles, as you say you can take information from when you restart (or die) into the next go so its infact the same

Happy hunting.  

[THobson]

Here is the analysis for the Old Base and the Ammo & UAZ.  Really quite different.

        Old Base     Ammo & UAZ
Loons        %             %   
0              0.6           0.9   
1              0              0   
2              0.03         0.7   
3              0.3           2.5   
4              1.1           3.9   
5              2.3           4.2   
6              2.8           5.0   
7              2.1           7.3   
8              1.7           9.8   
9              2.9           10.7   
10            5.4           10.2   
11            7.9           9.6   
12            8.8           9.3   
13            7.8           8.6   
14            6.3           7.1   
15            6.1           5.0   
16            7.2           2.9   
17            8.7           1.4   
18            9.1           0.6   
19            7.9           0.2   
20            5.6           0.06   
21            3.2           0.01   
22            1.4           0.0023   
23            0.5           0.0002   
24            0.14         0   
25            0.0232     0   
26            0.0036     0   
27            0.0001     0
   
Expected numbers:
OldBase        14.2
Ammo & UAZ  9.9   

[THobson]

macguba:

Here is the C program I did the simulation with.  I am sending it to you for two reasons.  

1. One you might find a C complier (it really is written in C and not C++ so free compilers are easily found) and you may wish to use it for analysing later versions of the mission yourself.  It is a simple matter to change.

2. If you want  you can update the probabilities in the program with those for your latest version and send it back to me.  I will happily run the program for you and send you the results.

The file is just a normal text file that can be opened and edited in Notepad.  The .c extension just tells the compiler that it is a C program.

The probabilities are stored in two arrays.
GrpProb contains the probability of each group existing.  For example:
GrpProb[1] = 90;
means that Group 1 has a 90% probability of existing

LoonProb contains the probability of each loon existing if the group exists.  For example:
LoonProb[1][6] = 32;
means that the probability of Loon6 in Group1 existing is 32% if Group1 exists.  Change the [1] to [2] and you get the probabilities for Group2, and similarly for Group0

(note I have numbered the groups from 0 to 2 and loons from 0 to 8.)

There is a section in the Initialisation procedure for each of the possible starting locations,  The comments should clearly identify where.

In case you ever do run the program the line near the top:

int StartPoint = infil1;

tells the program which of your locations to do the analysis for  (I followed your marker names and used infil0 for the Ammo & UAZ location).  So you only need to change this line to refer to the location you want to analyse.

The output format is ugly but works.  It takes 4 to 5 seconds to simulate 1 million mission starts.

Anyway - if you want to update the probabilities and send it back to me I will send you the analysis.

I tried to send this by PM but could find no way to include an attachement.

[macguba]

These are very interesting numbers.   My immediate reaction was to do some test runs.

Bizons for everybody.    Hit the dirt defence.   Play till the end of the initial attack.   When the other start postiions came up I just hit retry.

Old Base

1) died - bizon    (1 casualty before I went down)
2) no casualties
3) no casualties
4) 1 loon injured

Ammo & UAZ

1) died - M16 (I was first casualty)
2) no casualties
3) no casualties
4) no casualties (but this is the third time in a row the rear attack group (70%) has not appeared)


BMP Ambulance & BRDM

1) died - G36 (first casualty)
2) 4 casualties.  never saw the enemy myself
3) died  (2 known casualties when I went down)
4) no casualties


This is not a big enough sample to draw any firm conclusions but my findings are

1) I take a couple of goes to warm up

2) I've played this mission before

3) The 70% probs are the ammo & UAZ are a mistake - they should be 75% like the others.   I mean I probably missed them out when increasing everybody else to 75%, not being on the circle that start position does get forgotten sometimes.

4) BMP Ambulance is still too hard because most of the attacks are from behind where the player can't help.


(Edit:  just seen your next post)

(Edit edit:  and the previous ones.  ;D)

Just found something called lcc-win32 which claims to be a C compliler.  Now I just how to figure out what to do to it.

Can't get that to work, found another one called Miracle C.   I can open the file unimpossible.c and click "compile" and it gives me an output window with this in it

Miracle C Compiler (r3.2), written by bts.
Compiling c:\documents and settings\duncan\desktop\un-impossible.c
Initialise
IntRand0_99
LoonsInGrp
main

19% cg space used
Time taken: 0.078 seconds

The trouble is I don't really know what I'm trying to do.    Is compiling the same as running it, or do I have to run it after it has been compiled?

Ah-ha, making progress ... compile, build and run.    All I have to do now is figure out how to stop the dos window with the output from vanishing as soon as it appears.

Bingo.  Run it in a dos window.    It's a long time since I typed cd\