Old 'fixed' bugs continue to re-appear

[Darth Ender]

Sally, don't know if it can be of any help, but one thing i'm noticing very often upon exit, is that the game shows for few second a frame, from several minutes before i actually exited the game. For example i exit the game while landed at a station and inside my ship. I exit the game and i'm presented with this still frame (if i was in the concourse before, it's an image from there, or it might be me approaching the station, it depends from what i was doing before exiting the game).

It looks like old frames do not get flushed out of the rendering pipeline or memory. Sorry for the non technical explanation, not that tech savvy

what they do in the rendering pipeline or not regarding old data wouldn't keep the game from closing like you are saying it is.

That sounds more like the client is waiting for a response from the server before exiting and it's staying in a state that was never expected to be something that is visible to the end user for very long if at all.

 

[Mr T]

Sometimes software development can be like a game of whack-a-mole, as the bugs keep re-appearing after you've squished them:

 

[Darth Ender]

Sometimes software development can be like a game of whack-a-mole, as the bugs keep re-appearing after you've squished them:

they don't reappear by magic.

they are reintroduced by the same people who supposedly squashed them due to negligence.

 

[Zeke Drexel]

I thought Odyssey was meant to be a ground up rebuild of ED? Hence all these teething issues ...

 

[GroG79]

Aaaaany instances of this continuing (and I'm sadly seeing a fair few right now) I am MASSIVELY interested in some like...more help nailing it down (so sorry to ask).
What I'm specifically looking for at the moment is anyone who might have exited after using the Suit extras shop at time of exiting or near to exiting. This is so tricky...

It might be totally wrong but it seems to not close the game for me if I have edmc running

 

[taniwha-qf]

they are reintroduced by the same people who supposedly squashed them due to negligence.

More often, the bugs were not fixed properly in the first place. That is, the root cause of the problem was not found and fixed, but rather a bandaid was applied and thus changes elsewhere bring the bug back out of hiding.

 

[sallymorganmoore]

Sally, don't know if it can be of any help, but one thing i'm noticing very often upon exit, is that the game shows for few second a frame, from several minutes before i actually exited the game. For example i exit the game while landed at a station and inside my ship. I exit the game and i'm presented with this still frame (if i was in the concourse before, it's an image from there, or it might be me approaching the station, it depends from what i was doing before exiting the game).

It looks like old frames do not get flushed out of the rendering pipeline or memory. Sorry for the non technical explanation, not that tech savvy

Something REALLY is hanging on in "processes" somewhere, right? I think we're getting closer. The team are massively grateful for all of this detective work!

 

[cmdr milesred]

I thought Odyssey was meant to be a ground up rebuild of ED? Hence all these teething issues ...

NO, not even.
EDO is a replica of the shell of Horizons, suits replaced ships, etc.
And for a lack of reality, pasted into the mix. hence the numerous disconnects.
the empty shell is what has been getting filled in by these patches.. and as mentioned before, it took 4 years to make EDH a smooth as it was.

If you were to lay out all the bugs and discussions for EDH, you would see a mirror resemblance of what we went through, now doing it again for EDO, because EDO came empty.
I will be at least another year older by the time EDO is anywhere near as complete and functional as Horizons was pre EDO. we are almost at the 1/2 way mark time-wise, and look where it sits...

If that rumor had any truth, it left with the alpha.
.

 

[Mogen Teras]

NO, not even.
EDO is a replica of the shell of Horizons, suits replaced ships, etc.
And for a lack of reality, pasted into the mix. hence the numerous disconnects.
the empty shell is what has been getting filled in by these patches.. and as mentioned before, it took 4 years to make EDH a smooth as it was.

If you were to lay out all the bugs and discussions for EDH, you would see a mirror resemblance of what we went through, now doing it again for EDO, because EDO came empty.
I will be at least another year older by the time EDO is anywhere near as complete and functional as Horizons was pre EDO. we are almost at the 1/2 way mark time-wise, and look where it sits...

If that rumor had any truth, it left with the alpha.
.

To be honest, I'm just as guilty as everyone else for fogetting how Horizons launched.

 

[Mr T]

they don't reappear by magic.

they are reintroduced by the same people who supposedly squashed them due to negligence.

What? There's no magic dust holding it all together?

I could be wrong but I just think the game is so complex and there are so many inter dependencies that's it's near impossible for the DEVs to think of everything that may be affected by each change. It doesn't help that there's so many changes all happening at once either. I mean, look at the length of some of those patch notes.

 

[cmdr milesred]

no program is so complex that a proper flowchart can't be made for it, along with a storyboard for the path, along with someone directing it's progression.
A game or any professionally created program is similar to making a tv show or a cartoon. end result should not be by accident.
programmers don't have to know everything, only how to code efficiently, effectively and howto document any/all changes/additions they do.
Like a welder or the maintenance people or the electrician adding the lights.

ever see a logic equation for creating a chip? that's math. math that some people(very few) can do in their heads. the more complex, the more documentation and people paying close attention are needed to ensure it is acceptable.
most require flowchart and simulator and the book that created it(documentation) to follow.

The problems are many, number one being, its people. people are not machines and we make mistakes, we forget things, sometimes faster than we can write them down.
its common to create a routine in your head, implement it, then spend hours or days or weeks documenting it. god forbid you have another brilliant idea in the midst of this...and on and on it goes.

 

[taniwha-qf]

Yup, a flowchart can be made for any program no matter how complex it is (otherwise, optimizing compilers wouldn't have a chance at anything more complex than peephole and jump-threading optimizations). However, those flowcharts will be horrifically complex. Sure, you can reduce that complexity, but by the time you get it down to something manageable (assuming that's possible), you've lost the details on the interactions between different parts.

Despite the name "computer science", software is not science, nor engineering, and not even art. Due to the inherent emergent behavior, software is, essentially, magic. And even the tiniest mistakes can bring out the dragons. That we've managed to tame even a tiny portion of that magic is amazing.

And math is not infallible. 0.999... == 1? It depends (but usually). 1+2+3+... == -1/12? It depends. sin(x)/x for x=0? It depends on how you calculate it (use L'Hopital or Taylor series expansion, you get 1, though). Will 3n+1 for positive integers always get you to 1? Unknown, but it does for all values up to 2^68. Even an RS latch, the cornerstone of logic machines, has undefined behavior for one input state.

But, here's a little math for you: every conditional in a program doubles the complexity of that program. I'm pretty sure there are more than 64 conditionals in Odyssey. Have fun with that flowchart.

 

[Brrokk]

Yup, a flowchart can be made for any program no matter how complex it is (otherwise, optimizing compilers wouldn't have a chance at anything more complex than peephole and jump-threading optimizations). However, those flowcharts will be horrifically complex. Sure, you can reduce that complexity, but by the time you get it down to something manageable (assuming that's possible), you've lost the details on the interactions between different parts.

Despite the name "computer science", software is not science, nor engineering, and not even art. Due to the inherent emergent behavior, software is, essentially, magic. And even the tiniest mistakes can bring out the dragons. That we've managed to tame even a tiny portion of that magic is amazing.

And math is not infallible. 0.999... == 1? It depends (but usually). 1+2+3+... == -1/12? It depends. sin(x)/x for x=0? It depends on how you calculate it (use L'Hopital or Taylor series expansion, you get 1, though). Will 3n+1 for positive integers always get you to 1? Unknown, but it does for all values up to 2^68. Even an RS latch, the cornerstone of logic machines, has undefined behavior for one input state.

But, here's a little math for you: every conditional in a program doubles the complexity of that program. I'm pretty sure there are more than 64 conditionals in Odyssey. Have fun with that flowchart.

I agree with your description of computer science but I think you're doing maths down a bit. 0.999.… equals 1 ; this doesn't depend on anything else once you've properly defined what you mean by the three dots (and if you haven't done that, you can't use maths on them). 1+2+3 doesn't equal -1/12; the operations leading to that conclusion contain error. sin(x)/x has no value at x=0, but it has a limiting value there which is unambiguously 0.

Off my hobbyhorse now.

 

[taniwha-qf]

I agree with your description of computer science but I think you're doing maths down a bit. 0.999.… equals 1 ; this doesn't depend on anything else once you've properly defined what you mean by the three dots (and if you haven't done that, you can't use maths on them). 1+2+3 doesn't equal -1/12; the operations leading to that conclusion contain error. sin(x)/x has no value at x=0, but it has a limiting value there which is unambiguously 0.

Off my hobbyhorse now.

I can't actually disagree with anything you said, but neither can I agree with it. Thus why I said what I said about math (and as for 0.999..., 1-Ɛ where 0 < Ɛ < R (optionally, Ɛ^2 = 0))

 

[Brrokk]

I can't actually disagree with anything you said, but neither can I agree with it. Thus why I said what I said about math (and as for 0.999..., 1-Ɛ where 0 < Ɛ < R (optionally, Ɛ^2 = 0))

Consider 1 - 0.999...
This can't be negative.
Therefore it's either 0 or positive.
If it's positive, define E=1-0.999... Because this is positive we can take S=-log10(E)
Define a number T which is 0. followed by at least S zeros, then a 1.
This is less than E but still greater than 1-0.999... which leads to a contradiction.
Therefore E can't be positive.
Therefore E=0, which means that 0.999... = 1 .

I've left many details out; this is the best I can do in a forum.

 

[taniwha-qf]

I've seen (and accept) the arguments for 0.999... = 1, but the (math) person making that argument conceded that accepting infinitesimals complicates things. 0.999... = 1 is the accepted standard because it usually makes things a lot easier. (iirc, it was a 3b1b episode)

 

[Brrokk]

I've seen (and accept) the arguments for 0.999... = 1, but the (math) person making that argument conceded that accepting infinitesimals complicates things. 0.999... = 1 is the accepted standard because it usually makes things a lot easier. (iirc, it was a 3b1b episode)

Sorry, but as I said this is my hobbyhorse.
No infinitesimals are involved. The difference 1-0.999... must be positive, negative or zero. It clearly can't be negative and assuming it's positive leads to a contradiction. Therefore it's zero, which means the two numbers are equal.

<Edit> Someone has just told me a nice illustration in a maths group. Divide 1 by 3. The answer is 0.333... Then multiply by 3.

 

[Deleted member 182079]

What? There's no magic dust holding it all together?

I could be wrong but I just think the game is so complex and there are so many inter dependencies that's it's near impossible for the DEVs to think of everything that may be affected by each change. It doesn't help that there's so many changes all happening at once either. I mean, look at the length of some of those patch notes.

I reckon it's more to do with

Poor documentation (be that within the code or outside of it)
Poor knowledge retention (staff turnover with insufficient knowledge transfer)
Disorganised org structure (one team not knowing what the other does)
Poor version control - the amount of regression issues is truly shocking and I can't think of another game that has so many of them
"that'll do" / "good enough" attitude with a tendency to put band aids on issues rather than fixing the root cause
Quality is a low priority within the organisation (look at the paintjobs if you want a simple example, and unlike other bits in the game those are an actual income stream)
Skill levels? I don't want to step on anyone's feet but FDev don't apparently pay that well, and this will likely have an impact on the type of people they can recruit.

 

[JonesTheFilm]

Spoiler: So say we all

 

[taniwha-qf]

<Edit> Someone has just told me a nice illustration in a maths group. Divide 1 by 3. The answer is 0.333... Then multiply by 3.

This shows only that decimal representation is only an approximation of fractions ie, 0.333... is not exactly 1/3, but infinitesimally shy of 1/3. Thus, we just don't worry about it and define that 0.333... is 1/3, and 0.999... is 1. It'll take a while for the error to cause problems