Wasteland People Bitching About Jamming

[toro]

Terrible optimization? That shit can fetch one integer and add it to another up to THREE TIMES per someone shooting a gun. Hello framerate killer!

Premature optimization is the root of all evil.

What happens when there is no other way to implement something except ... in a optimized way? Like for embedded systems, automotive, aviation, space stuff?

 

[Norfleet]

What happens when there is no other way to implement something except ... in a optimized way? Like for embedded systems, automotive, aviation, space stuff?

Then it wouldn't be premature, now would it?

 

[toro]

What happens when there is no other way to implement something except ... in a optimized way? Like for embedded systems, automotive, aviation, space stuff?

Then it wouldn't be premature, now would it?

Depends how you define premature. But don't let semantics stand in your way

 

[tuluse]

What happens when there is no other way to implement something except ... in a optimized way? Like for embedded systems, automotive, aviation, space stuff?

The key word is premature.

Edit: oops

 

[Jasede]

lol jamming

The best guns have 4% jamming chance
You can get the -4% jamming chance items at level 1

I'm on SJ and nothing has ever jammed for me, only for enemies - on account of my jam chance being 0%. If anything the jamming needs to be more frequent.

 

[toro]

What happens when there is no other way to implement something except ... in a optimized way? Like for embedded systems, automotive, aviation, space stuff?

The key word is premature.

Edit: oops

The key word is not applicable to certain software projects (embedded/HW drivers) because usually either the code is optimized and it works or it doesn't work at all. There is no proper time to do optimizations or not.

 

[Shadenuat]

lol jamming

The best guns have 4% jamming chance
You can get the -4% jamming chance items at level 1

I'm on SJ and nothing has ever jammed for me, only for enemies - on account of my jam chance being 0%. If anything the jamming needs to be more frequent.

except only rifles of both types and pistols can get magazines mods. well, lazors don't jam too.
great fun running in front of cluster of mobs with Jackhammer, jam, cry cock, use unspent action points to run all way back before you get raepd

 

[Jasede]

I don't mean to sound like a combat-fag, but it'd never occur to someone like me (or Mondblut) to ever use weapons that aren't statistically the best. In this case that'd mean either 4x Assault Rifles or 3x Assault Rifles 1x Sniper/GammaRay.

Unless you're a casual and don't play SJ, then you can pick whatever weapon you want since the game was apparently balanced for The New Generation.

 

[Shadenuat]

I don't mean to sound like a combat-fag, but it'd never occur to someone like me (or Mondblut) to ever use weapons that aren't statistically the best. In this case that'd mean either 4x Assault Rifles or 3x Assault Rifles 1x Sniper/GammaRay.

Unless you're a casual and don't play SJ, then you can pick whatever weapon you want since the game was apparently balanced for The New Generation.

So... weapons that are suboptimal and can't be upgraded are alright because you are too good of a player to use them?


Also, Jackhammer is a fine weapon, it's 3x burst like rifle but in cone. You don't even need 4-5 assault rifles to play on highest difficulty, 2 suffice. Only melee/heavy/pistols get short end on SJ.
Well, heavy/pistols just suck.

 

[Jasede]

You are correct, the Jackhammer is pretty good.

 

[tuluse]

Pistols are good for spreading damage to multiple targets because of the low AP cost. Great for finishing off when enemies are super low on health.

 

[Spockrock]

I would really like to see you stumbling in explaining the O(n) shit.

 

[toro]

I would really like to see you stumbling in explaining the O(n) shit.

Nice cover.

 

[toro]

Sorry for the late reply.

Basically I don't see anything wrong with the code but we don't have all the code in that picture: please post the content of current.GetBonus("chanceToJam") function.

Until then my assumption is that Unity's RNG generator might be fucked meaning that it generates more numbers closed to 0 (could be a special case just for some type of processors).

Other than this, the code is unoptimized: the addition of chanceToJam each fucking time is retarded. I wonder why?

What is wrong with calculating each chance by itself? That's what random means.

There is nothing wrong with checking the chance-to-jam for each shoot, in fact that's the purpose of that code and what is needed.

The problem is that the code is calculating the chance-to-jam threshold every-time a player/npc is shooting -> the threshold is the return of GetChanceToJam() function.

There is no need to compute the chance-to-jam each time you shoot because this threshold doesn't change between shoots (unless inXile implements weapon degradation).

And it's obvious that a sane implementation would save this threshold per avatar/npc and then update it only when change conditions are present: weapon was changed, new mod was applied or removed and so on.

But I guess you want them to use a table that is filled in advance, so it is "fair" and jams exactly 2 times in every 100 shots, if it's 2%?

This is actually the point of the entire discussion. Maybe there is a confusion on my side or not but imweasel stated something like that: "5% chance to jam ≙ statiscally, 1 out of 20 shots will cause your gun to jam"

The fact is that the implemented code has nothing to do with what imweasel stated cause it just does the following:
- compute chance-to-jam threshold, which is redundant in my opinion irrelevant of run-time penalties,
- takes a random number in the 0 to 100 range,
- compares this to the threshold and return true/false depending on the comparison result.

It is pretty evident that this entire implementation depends on the RNG quality and that it doesn't implement a statistical model.

I have no strong personal preferences but a statistical model might be better for one reason -> reducing save scumming.

Basically with the current implementation, as long as you have 1% chance of opening a safe then you can save scum until you will open that safe.

But if a statistical model system was implemented like imweasel said then it would be impossible to open some safes because the model will dictate that you must have 2% percent of failures.

How this statistical mode is implemented is not really important. It can be done with lists or math formulas considering the past success rate or other factors. It really depends what the developer wants.

 

[gromit]

There is no need to compute-and-store the threshold on weapon / swap. There's no need to add calls to recalculate in every "and so on." There's no need to introduce the potential for any of those numbers to ever fall out of sync, for any reason, no matter how "wrong" the code that causes that to happen would be.

The code adds four numbers together each time you shoot, in a turn-based game.

A "sane implementation" of anything would avoid creating any complication, or adding any work, no matter how minor, for a benefit that would be barely noticeable... even in ideal conditions and under massive workload.

I gave you a very trustingly-written bit of theory which you chose to ignore.

So look, here: will this make you happy?

toro has correctly identified three numbers which could be summed prior to that point

 

[toro]

There is no need to compute-and-store the threshold on weapon / swap. There's no need to add calls to recalculate in every "and so on." There's no need to introduce the potential for any of those numbers to ever fall out of sync, for any reason, no matter how "wrong" the code that causes that to happen would be.

The code adds four numbers together each time you shoot, in a turn-based game.

A "sane implementation" of anything would avoid creating any complication, or adding any work, no matter how minor, for a benefit that would be barely noticeable... even in ideal conditions and under massive workload.

I gave you a very trustingly-written bit of theory which you chose to ignore.

So look, here: will this make you happy?

toro has correctly identified three numbers which could be summed prior to that point

I did not see ur post. Sorry. I'm not happy.

 

[Immortal]

There is no need to compute the chance-to-jam each time you shoot because this threshold doesn't change between shoots (unless inXile implements weapon degradation).

Or if a player could spend AP to attach a mod in combat between gun shots..
What kind of horrible developers would design logic that accounted for that?

 

[Foobar]

ITF: people don't understand probability.

Most people, both designers, and users, see values of 90% or better as "very good". An engineer will tell you that even 99% is shit. Any engineer building you something that only works 99% of the time is going to find himself embroiled in a vicious lawsuit and/or lose his job.

It's true, my PCs tried to file a class action suit against negligent gun manufacturers. Sadly, most of those jerks (and the civil court system) got away on the technicality of being killed by the apocalypse.

TL;DR for this entire topic: Melee, Sturdy Mag, Energy. Done.

 

[Norfleet]

Or if a player could spend AP to attach a mod in combat between gun shots..
What kind of horrible developers would design logic that accounted for that?

Actually, I believe the issue in question stems from the fact that the function is called multiple times for the SAME gunshot. This is sort of a programming sin, since calling that blackbox function may entail numerous redundant computations on something that should not be changing anyway. A decent compiler would probably optimize that, albeit I would not call the .NET compiler very decent, if that kind of structure is appearing when you decompile it back. However, I don't really think this is important. It's a relatively small function where the redundant calls are only made very infrequently relative to the rest of the program, as the game is not exactly a thousand-round all-you-can-eat projectile buffet. It's bad coding, but it's not bad coding that affects anything important. There are certainly bigger fish to fry.

 

[Immortal]

Actually, I believe the issue in question stems from the fact that the function is called multiple times for the SAME gunshot.

Uhmmm.. Did you read the code?

 

[toro]

Actually, I believe the issue in question stems from the fact that the function is called multiple times for the SAME gunshot.

Uhmmm.. Did you read the code?

This is actually a comedy thread. Did you get the memo?