[1.3.1] Ferram Aerospace Research: v0.15.9.1 "Liepmann" 4/2/18

[Jacke]

57 minutes ago, MunGazer said:

I think it'd be cool if Squad came to ferram4, and said hey man, we really like your work, and we want to collaborate with you.  ....

Well....  A bunch of modders did get recruited by Squad during the development of KSP 1.2, not to incorporate their mods but to help make KSP 1.2 better.  And they did help make KSP 1.2 better.  And then about a week before 1.2 release, a bunch of those and other staff parted company with Squad.  We don't really know how or why.  So, I'm imagining any modder would be very cautious now about accepting any deals with Squad.

Despite recent improvements, I really depend on mods to turn KSP from a game that would frustrate me with the many things it lacks to a game I enjoy.  And for most of those required mods, many of them much simpler that FAR, there's no sign that stock KSP will ever incorporate them.  Or Squad has even said they'll won't entertain such a change.  As an example, the information from engineering and info & control mods, like the delta-V of a design and the remaining delta-V in flight, or the shape of the current orbit visible in flight mode, like apoapsis, periapsis, and inclination.  I think Squad looks at the big change they did with aero back in KSP 1.0 as sufficient and have no interest in going further with it.

[Beetlecat]

17 minutes ago, Jacke said:

Well....  A bunch of modders did get recruited by Squad during the development of KSP 1.2, not to incorporate their mods but to help make KSP 1.2 better.  And they did help make KSP 1.2 better.  And then about a week before 1.2 release, a bunch of those and other staff parted company with Squad.  We don't really know how or why.  So, I'm imagining any modder would be very cautious now about accepting any deals with Squad.

Despite recent improvements, I really depend on mods to turn KSP from a game that would frustrate me with the many things it lacks to a game I enjoy.  And for most of those required mods, many of them much simpler that FAR, there's no sign that stock KSP will ever incorporate them.  Or Squad has even said they'll won't entertain such a change.  As an example, the information from engineering and info & control mods, like the delta-V of a design and the remaining delta-V in flight, or the shape of the current orbit visible in flight mode, like apoapsis, periapsis, and inclination.  I think Squad looks at the big change they did with aero back in KSP 1.0 as sufficient and have no interest in going further with it.

And there are still active modders employed by Squad.

Edited December 9, 2016 by Beetlecat

[MunGazer]

2 hours ago, tetryds said:

I'm afraid that is not how money works.

No prob, I don't claim to know how money works specifically with Squad employment.  I just think that ferram4's work is nice enough to warrant something more than donations, but you could go so far as to say "MunGazer you are absolutely ignorant about how money works in this world, and you have no business talking about it".  Aight, no prob.  I just think it'd be great if we had a button in settings "switch to FAR" lol

[Drew Kerman]

Why anyone would want any mod to become stock and thus be locked into a months-long update cycle is beyond me. Just saying Looking forward to moving up to v1.2.2 soon^TM

[MunGazer]

1 hour ago, Drew Kerman said:

Why anyone would want any mod to become stock and thus be locked into a months-long update cycle is beyond me. Just saying Looking forward to moving up to v1.2.2 soon^TM

I didn't know adding it in would drastically draw out updates.  I thought maybe if the code got properly factored in then the updates to FAR could be a little more streamlined/dynamic.  But what do I know, I'm not a dev :P.  Plus, it might actually be nice to have a collection of mods be persistent for a while as opposed to constantly falling behind updates.

[steve_v]

5 hours ago, MunGazer said:

add your aerodynamics to our game

The other problem with this would be the licence...  FAR is GPL so you can't just incorporate it into KSP, it'd have to stay as a plugin, the source would have to remain open etc. None of this fits with SQUADs commercial licence at all.
It'd be more like "come write a FAR-like aero system (without nicking any code from FAR)". Stock aero is good enough for most (no, I'm not "most" either ), so it'll never happen.

Edited December 9, 2016 by steve_v

[Drew Kerman]

2 hours ago, MunGazer said:

I didn't know adding it in would drastically draw out updates

s'ok yea most people don't realize, especially in this case since FAR is a lot more of a mature mod and also complex so it doesn't undergo a lot of frequent updates. But still, if you look at the period between Squad releasing KSP v1.1.2 and v1.1.3 Ferram pushed out two additional sizeable releases besides the compatibility updates that addressed some important issues. If it had been stock, ppl would be waiting until v1.1.3 of KSP to get any fixes to problems in FAR that came along with the KSP v1.1.2 release.

Edited December 9, 2016 by Drew Kerman

[Sol Invictus]

3 hours ago, steve_v said:

The other problem with this would be the licence...  FAR is GPL so you can't just incorporate it into KSP, it'd have to stay as a plugin, the source would have to remain open etc. None of this fits with SQUADs commercial licence at all.
It'd be more like "come write a FAR-like aero system (without nicking any code from FAR)". Stock aero is good enough for most (no, I'm not "most" either ), so it'll never happen.

Hey, it's not like ferram can't change the licence if paid sufficiently by Squad.

[kokospl]

7 hours ago, Jacke said:

As an example, the information from engineering and info & control mods, like the delta-V of a design and the remaining delta-V in flight, or the shape of the current orbit visible in flight mode, like apoapsis, periapsis, and inclination.

Why would anyone buy KSP 2.0, if the current version has all the things they need?

[steve_v]

2 hours ago, Sol Invictus said:

Hey, it's not like ferram can't change the licence if paid sufficiently by Squad.

Legally, I doubt he can. There are other contributors, so no one person owns all the code. The GPL is specifically designed to prevent this kind of malarkey.

[blu3wolf]

The collaborative nature of it would be the only issue. If all the contributors agreed, there would be no issue with them reissuing the code under a different license. That seems unlikely to happen, though.

Probably for the best. As much as FAR is an integral part of KSP to some of us, to many it is an unwelcome intrusion of the real world on their game mechanics.

[sarbian]

17 hours ago, Inf7nite said:

from what I know it is because you spend most of your time in space, not in air, so they find its not worth the effort.

Yeah sure. And that's why aero was reworked in 1.0 right ? Don't pretend you know anything about this.

[Sol Invictus]

1 hour ago, steve_v said:

Legally, I doubt he can. There are other contributors, so no one person owns all the code. The GPL is specifically designed to prevent this kind of malarkey.

If I understand it correctly, GPL makes it so that the piece of code licensed under it stays forever free. It doesn't prohibit mod authors from issuing code under different licence, for instance BSD. If ferram & co. did so, then we would have two identical versions of FAR, one licensed under GPL and another under BSD. The one licensed under GPL would remain as GPL indefinitely, and independently from it's BSD equivalent.

Edited December 9, 2016 by Sol Invictus

[tetryds]

There would be no point in simply throwing FAR at KSP stock as a mod, if it was to be implemented like that the best option would be a full reimplementation of it, deeper into ksp's code and allowing for better optimizations. That is why the license does not matter.

But it's a bit late for this discussion, like, more than a year late.

I personaly like stock aero as it is, it fits its role wonderfuly well. FAR should be meant for people wanting more, and that is what it is.

[Darael]

2 hours ago, Sol Invictus said:

If I understand it correctly, GPL makes it so that the piece of code licensed under it stays forever free. It doesn't prohibit mod authors from issuing code under different licence, for instance BSD. If ferram & co. did so, then we would have two identical versions of FAR, one licensed under GPL and another under BSD. The one licensed under GPL would remain as GPL indefinitely, and independently from it's BSD equivalent.

Correct (ish) - but it's the "& co" that's crucial here:  Part of the point of the GNU GPL is that every contributor would have to agree to do this, or all code touched by those contributors who did not explicitly so agree would have to be stripped out of the BSD licence.

And it'd probably have to be replaced by still other contributors who were given only a description of what the missing code did and had never seen the original, to ensure that the replacement wasn't derived from the GPL-only code.

As the number of historical contributors rises and their contributions gain dependencies in other parts of the code, the difficulty of a) contacting them and b) replacing code written by people who refuse or cannot be contacted both increase.

In short: Ferram alone can't change the licence.  Ferram et al can change the licence in theory, but it'd be a lot of work in practice.

(also there's no need to consider the differently-licensed versions distinct until and unless there are differences between them besides licensing, such as newly-contributed patches that are GPL-only (BSD-only ones can be relicensed as GPL by the terms of the BSD licence, so that way round is a non-issue).  So there'd still be one FAR, and it'd be dual-licensed, until and unless there was a subsequent single-licence release.)

[Ben Garner]

11 hours ago, Darael said:

Correct (ish) - but it's the "& co" that's crucial here:  Part of the point of the GNU GPL is that every contributor would have to agree to do this, or all code touched by those contributors who did not explicitly so agree would have to be stripped out of the BSD licence.

And it'd probably have to be replaced by still other contributors who were given only a description of what the missing code did and had never seen the original, to ensure that the replacement wasn't derived from the GPL-only code.

As the number of historical contributors rises and their contributions gain dependencies in other parts of the code, the difficulty of a) contacting them and b) replacing code written by people who refuse or cannot be contacted both increase.

In short: Ferram alone can't change the licence.  Ferram et al can change the licence in theory, but it'd be a lot of work in practice.

(also there's no need to consider the differently-licensed versions distinct until and unless there are differences between them besides licensing, such as newly-contributed patches that are GPL-only (BSD-only ones can be relicensed as GPL by the terms of the BSD licence, so that way round is a non-issue).  So there'd still be one FAR, and it'd be dual-licensed, until and unless there was a subsequent single-licence release.)

Do we know how much of the codebase includes contributions from contributors other than @ferram4? The 'clean room' approach might be feasible in practice if the contributions of others are not a huge portion. I think that there are likely enough people who would be both capable of and interested in this undertaking. As for myself, I do not currently have the required skills or knowledge to contribute to it, but I am actively trying to learn C# and how the KSP API works (I can do basic Java, so I'm not completely starting from scratch  ). The biggest problem would be that there are far fewer of us with a thorough understanding of aerodynamics, especially at supersonic speeds. Depending upon how specific the spec is allowed to be that's created from the original code, it could be more difficult than we might anticipate. Taking airflow equations and converting them each into a computer algorithm is no big deal, but actually understanding how to use them in concert to make air act like air is a whole different animal.

[ferram4]

Just to shut up this whole debate about whether or not FAR could have been / can be implemented:

Back before the aero overhaul happened for stock, I offered to help Squad out with it.  This was prior to the voxel method was developed and licensed and I likely would have tried to get that implemented as the system since it would be far easier to do in that system with the ability to top-down declare what meshes and colliders need to do and make things faster.  They politely declined, the voxel method got implemented into FAR under the GPL license, and here we are.  Also, as @tetryds said, a year late.

As for re-licensing, it is possible, and I do have a list of who did what so that I know what can be stripped out and replaced if necessary.  Most of the stuff from other people is actually stuff that should be replaced at some point with more accurate and faster code anyway, so there's very little issue with any of that.  The older versions of FAR's code would still be available under GPL, but the rest can run something else.

Now, behave while I try to figure out all the remaining things wrong with this monster.  Every time I look here there are multiple moderator posts about cleaning things up.  I don't want the thread locked (especially because it'll be a pain when I finally go to update) so don't give the moderators reason to do so.

[Nnimrod]

3 hours ago, ferram4 said:

Just to shut up this whole debate about whether or not FAR could have been / can be implemented:

Back before the aero overhaul happened for stock, I offered to help Squad out with it.  This was prior to the voxel method was developed and licensed and I likely would have tried to get that implemented as the system since it would be far easier to do in that system with the ability to top-down declare what meshes and colliders need to do and make things faster.  They politely declined, the voxel method got implemented into FAR under the GPL license, and here we are.  Also, as @tetryds said, a year late.

As for re-licensing, it is possible, and I do have a list of who did what so that I know what can be stripped out and replaced if necessary.  Most of the stuff from other people is actually stuff that should be replaced at some point with more accurate and faster code anyway, so there's very little issue with any of that.  The older versions of FAR's code would still be available under GPL, but the rest can run something else.

Now, behave while I try to figure out all the remaining things wrong with this monster.  Every time I look here there are multiple moderator posts about cleaning things up.  I don't want the thread locked (especially because it'll be a pain when I finally go to update) so don't give the moderators reason to do so.

we love you m8

[Flashblade]

5 hours ago, ferram4 said:

Back before the aero overhaul happened for stock, I offered to help Squad out with it.  This was prior to the voxel method was developed and licensed and I likely would have tried to get that implemented as the system since it would be far easier to do in that system with the ability to top-down declare what meshes and colliders need to do and make things faster.  They politely declined, the voxel method got implemented into FAR under the GPL license, and here we are.  Also, as @tetryds said, a year late.

That is really such a shame that this happened. I thought that Squad should at least implement something like your voxel based system themselves. Physics that act on an object as a whole compared to a collection of individual parts should be way easier on the cpu. Also that you don't need to declare cargo bays and fairings through entries in config files is just so much better, in fact that you could build bays and fairings out of individual parts yourself if you wished to do so.

I only think stock aero has really one thing up on FAR and that is landing and take-off speed for planes or i guess low-speed performance in general. In stock you can generally take-off and land with a speed that is approximately the same as in reality. The same plane can't be landed with a speed of about 80-90 m/s in FAR compared to stock where this is possible. In FAR this plane has to have a speed of at least 140 m/s or I can't stay on the necessary 2-3 meters vertical speed. This is insanely fast by comparison. This has probably something to do with the much higher weight of Kerbal crafts in general. I have just now started experimenting with secondary control surfaces but on the test craft that I outfitted with flaps and slats I did not see any noticible improvement on take-off speed, but I am a novice at this stuff and I might be doing it wrong.

[kcs123]

15 minutes ago, Flashblade said:

I only think stock aero has really one thing up on FAR and that is landing and take-off speed for planes or i guess low-speed performance in general. In stock you can generally take-off and land with a speed that is approximately the same as in reality. The same plane can't be landed with a speed of about 80-90 m/s in FAR compared to stock where this is possible. In FAR this plane has to have a speed of at least 140 m/s or I can't stay on the necessary 2-3 meters vertical speed. This is insanely fast by comparison. This has probably something to do with the much higher weight of Kerbal crafts in general. I have just now started experimenting with secondary control surfaces but on the test craft that I outfitted with flaps and slats I did not see any noticible improvement on take-off speed, but I am a novice at this stuff and I might be doing it wrong.

Take off/landing speeds would be much, much better once ground effect is put in consideration. In stock aero, low speed handling is same on the ground and 1km above ground. It should not be like this, but stock aero does not simulate ground effect either. Once aircraft is in air, it should require much more velocity to be able build up enough preassure under wings for lift, like it is in FAR.

On the other hand, while waiting for FAR to update, I tried to recreate craft with variable geometry wings from KSP 1.1.x release created in FAR.
Few control surfaces were needed to re-attach due to KSP/mods changes, but on my suprise, craft is handled quite close in stock aero as it was in FAR. Although, take off speed in stock is about 10 to 20 m/s lower than it was in FAR, as much as I recall it. I didn't actualy measure behaviour and making proper comparison.

[cantab]

Yeah, getting a plane with a nice slow landing speed is tough in FAR, a problem compounded by the desire to land in the bush and the additional difficulty of wheels now.

But I think this brings up what I see as the difference in philosophy between stock aero and FAR. Stock aero is happy to fiddle things to get a desired 'feel'. Landing speeds too high? Just tweak the lift behaviour to bring them down! FAR strives for objectively realistic behaviour. Landing speeds too high? Not FAR's problem provided the aerodynamics is correct for the craft you built. Rather look for ways that craft is unlike real aircraft, and like Flashblade said it's probably down to weight, an entire Cessna 172 with fuel weighs about as much as a Mk1 cockpit alone.

[kraden]

4 hours ago, cantab said:

an entire Cessna 172 with fuel weighs about as much as a Mk1 cockpit alone.

Every time I go to build a plane in KSP I think of this, lol (My father-in-law has a 172). Angle of attack should help take-off and landing speeds, at least a little bit.

[FunnelVortex]

The high takeoff and landing speeds actually are a problem I have with FAR (still love the mod). 

I can't even seem to even get wide-winged stock aircraft like the Stearwing A300 off the ground or landed with a reasonable takeoff/landing speed with FAR installed, despite the fact those wings should give it a lot of lift, especially at lower speeds and ground level.

Hopefully future versions of FAR will address this.

Edited December 10, 2016 by FunnelVortex

[tetryds]

@FunnelVortex as @kcs123 mentioned, FAR does not currently simulate ground effect, which is the main reason why takeoff and land speeds are higher than expected.

But, as @cantab said, stock crafts are a lot heavier than their real life counterparts, looking alike is far from behaving alike, especially when dealing with different masses and weight distribution.

It is expected that FAR has a bit lower low speed drag than it should, but that was more deeply tested a while ago by NathanKell as he created a cesna 172 as realistically as possible, many things changed since then

The long-awaited-and-still-waiting wing overhaul on FAR should address that, but no one knows when/if that will happen.

For now you can easily get away making proper use of flaps and spoilers, balancing your craft and making it lighter.

[FunnelVortex]

21 minutes ago, tetryds said:

@FunnelVortex as @kcs123 mentioned, FAR does not currently simulate ground effect, which is the main reason why takeoff and land speeds are higher than expected.

But, as @cantab said, stock crafts are a lot heavier than their real life counterparts, looking alike is far from behaving alike, especially when dealing with different masses and weight distribution.

It is expected that FAR has a bit lower low speed drag than it should, but that was more deeply tested a while ago by NathanKell as he created a cesna 172 as realistically as possible, many things changed since then

The long-awaited-and-still-waiting wing overhaul on FAR should address that, but no one knows when/if that will happen.

For now you can easily get away making proper use of flaps and spoilers, balancing your craft and making it lighter.

 

I'm personally hoping that is coming this next update, and turns out that is why it's taking so long .