Star Citizen: Death of a Space Thread?

The Something Awful Forums > Archives > Comedy Goldmine > The Goodmine > Star Citizen: Death of a Space Thread?

«‹›6530 »

DancingShade: Jul 26, 2007; by Fluffdaddy

Raskolnikov posted:

Could have fooled me. *tips fedora*

Welcome to SA by the way. :tipshat:

# ? Dec 29, 2016 10:08

Adbot: ADBOT LOVES YOU

# ? May 9, 2024 10:58

moveable shape: Oct 18, 2015

Sarsapariller posted:

Fundamental Truths About Star Citizen
1) $140 million and 5 years are not unreasonable numbers because they are developing new technology to realize a game beyond the scope of any other. Here, look at this new feature.
2) This new feature doesn't indicate any kind of dev time or money spent because it's built into the engine. Chris Roberts is an excellent steward of our money.
3) This new feature is our fault because they have to keep selling us stuff to produce the game- $140 million isn't enough to fund all these new features.
4) They've spent the last 4 years modifying the engine to the point that they completely rebranded it and had to hire all of the Crytek developers to come work on it.
5) The engine is still baseline enough that it can be swapped out for a completely different branch with no issues.
6) Most of the work of the last two years was on netcode and back end support, which is now complete, allowing the game to expand much faster.
7) Swapping to the new engine allows them to use Amazon's net code which is much better than theirs, and will allow the game to expand much faster.
8) What's in the game today is amazing, and if they were to stop development and just polish that and add some maps, they could probably release it there.
9) The current gameplay is clearly a pre-pre-alpha and fundamentally does not represent the final state of the game. Bugs should be expected.

nailed it

# ? Dec 29, 2016 10:11

Ghostlight: Sep 25, 2009; maybe for one second you can pause; try to step into another person's perspective, and understand that a watermelon is cursing me

Qwertycoatl posted:

if CIG are lying to exaggerate their progress

hosed up if true

# ? Dec 29, 2016 10:20

Skellybones: May 31, 2011; Fun Shoe

Tippis posted:

It's certainly �crazy�. It's also meaningless, and as such explains why they managed to do it in 1�2 days and why they're making such a big deal out of it. CIG has made a business out of delivering nothing � it's only fitting that their big engine upgrades consists of doing absolutely nothing.

Exactly, all that and it's free!

# ? Dec 29, 2016 10:33

no_recall: Aug 17, 2015; Lipstick Apathy

Sabreseven posted:

I had to check, it's not actually a photoshop, it's on the site clear as day.

I wonder how they worked that figure out, it's in linear kilometres and not 'squared' or even 'cubed', yet how do they make such a claim when they know full well 99.9 of those 100 are full of empty space and nothing to find?

I'm leaning towards thinking "false advertising".

1 x sextillion km = 1,000,000,000,000,000,000,000km = no way they made an in game map 100x larger than that.

105,700,083 light years?

They made the biggest gameworld yet!

Buy an Idris!

# ? Dec 29, 2016 10:47

Baxta: Feb 18, 2004; Needs More Pirate

ewe2 posted:

Sorry in advance for this rant, I just wanted to make it clear as clear why there are problems with CIGs/Parrys "explanation". Please point out mistakes or confusions/omissions.

Just managing the pieces of a website in git can teach you a lot about how VCS actually works and not how Ben Parry wants you to believe, like just skipping main development line history. Good VCS patterns are about minimising merge conflict and small, manageable deltas, regardless of the size of the project.

Let me break it down into a hypothetical, the kind of hypothetical that the Reddit crowd hates because they'd have to apply braincells to it, and let's pretend we're using git because its an excellent distributed VCS that lets you decide how to manage your workflow:

You've got engine code, it's made up of different bits that different groups of developers can work on, and like a good game company you want to make modifications to make the greatest game ever and you want builds to work so you can demonstrate how things are going (you know, not like SC). There will be a bunch of stuff that doesn't use the engine at all, ignore that in this discussion, it would be added at the build branch stage to make a build work and then get added to the main branch. Here we're only concerned with bits that need the engine code.

Ideally we'll have a development branch, a build branch and a main branch. You're the guy coordinating this because your developers are all working on their bits of the engine and when they put in changes, you make sure all the different bits agree and you take a snapshot of the current development branch and call that the build branch (and add those bits I mentioned above), and if it builds, those changes make it into the main branch. Clear so far? You can give these snapshots whatever name/version you like as long as it makes sense to everyone.

Ok, so down the track your builds are going fine but meanwhile the engine developer has released a new version of the engine and you want to take advantage. So you grab the new engine code, and to be safe, you clone a new copy of the original engine code and you do a merge to see what the changes are and make a diff from that. You now literally have what changed between engine versions. Well, like you would expect with a new engine version, they've changed the API you use to access its code and they've restructured its different bits quite a bit, there are new bits, older bits have gone. You do the same technique as before, comparing your current main branch (NOT the development branch, guess why) to the code and oh dear there'll be merge conflicts everywhere, the engine is quite different. What to do?

At this point we are up to BEFORE the LY fork, at 3.8 whatever version. We can break the task down: some bits can be easily discarded, some bits will be added, but there'll be a chunk that will have to be integrated piece by piece preferably along structural lines and meanwhile you have to educate the developers who can't continue along the lines of continuous development/builds until these changes are made. You have to educate the developers on the changes to the API, you have to reassign developers to the bits that are changed/new, and you may have to take a deep breath and throw a lot of good code away simply because you don't have the time or resources to rebase everything. "Rebase" sounds simple but only works if your modifications are only in the parts of the new code that didn't change, and in the case of an engine only some of that will be true.

Ideally, you can give some developers the freedom to rebase their own development work on the new engine code, but in some cases you'll have to do the work yourself and get them to clone from that. And THEN it's back to making the builds work, having "rebased" on the new code and then you've got a main branch which is a fork off the engine code.

And THEN you have the LY fork. Take the above scenario and repeat, unless you're prepared to simply throw away all the previous work. Unless the LY fork is amazingly compatible which I do not believe for a second.

I want to point out that as far as I understand, game development is rarely this organized or as careful. An object hierarchy is fine for business software development but usually only gets as far as the UI and maybe file operations but everything else is written pedal to the metal good old functional programming and that is a dog to manage even with the help of a VCS. The whole point of object orientation was to make large projects even feasible but that breaks down with the way games work which do multiple things at the same time like a mini-OS, it's that complex.

So when Ben Parry tries to tell me that it was easy I'm just dumbfounded. Because he should know better. And this discussion is only applicable to if they did it as he claims, ripptide might be closer to the truth.

This is a good post and explains exactly why the only thing they could have possibly done is used LY for the 3.7 fork making the whole transition absolutely pointless (Unless they are stupid enough to think LY will somehow allow them to produce their insane promises)

As I am a petty child with too much disposable income, I keep my account to smash my ships into everyone at Olisar.

# ? Dec 29, 2016 10:49

moveable shape: Oct 18, 2015

Rad Russian posted:

All the shill streamers have distract material ready to go on secondary screen that they constantly switch to, in order to hide the fact the whole game needs to be rebooted after every match and then takes forever to load.

The reboot glitch is in the live build? Are you making GBS threads me?

# ? Dec 29, 2016 10:49

AP: Jul 12, 2004; One Ring to fool them all
One Ring to find them
One Ring to milk them all
and pockets fully line them; Grimey Drawer

The old and tired "when will the game come out" argument.

# ? Dec 29, 2016 11:01

Second Sun: Apr 6, 2009; by FactsAreUseless

no_recall posted:

105,700,083 light years?

They made the biggest gameworld yet!

Buy an Idris!

Our galaxy is only 100,000 ly across. What the hell are they trying to say?

# ? Dec 29, 2016 11:01

AP: Jul 12, 2004; One Ring to fool them all
One Ring to find them
One Ring to milk them all
and pockets fully line them; Grimey Drawer

https://www.youtube.com/watch?v=HEZYkYmb8V0&t=27s

# ? Dec 29, 2016 11:09

AP: Jul 12, 2004; One Ring to fool them all
One Ring to find them
One Ring to milk them all
and pockets fully line them; Grimey Drawer

https://www.youtube.com/watch?v=ybf_ZtFYXqY&t=6s

# ? Dec 29, 2016 11:12

Skellybones: May 31, 2011; Fun Shoe

moveable shape posted:

The reboot glitch is in the live build? Are you making GBS threads me?

You know how hard it is to get a multiplayer FPS working in Cryengine?

# ? Dec 29, 2016 11:13

AP: Jul 12, 2004; One Ring to fool them all
One Ring to find them
One Ring to milk them all
and pockets fully line them; Grimey Drawer

YOU HAVE TO LOOK UP TO GO UP THE LADDER OF THE 315P

STEPS TO REPRODUCE
1 Looking slightly down results in climbing through the ladder
2 Looking up results in climbing up the ladder

ACTUAL RESULT
Unless you are looking up you won't climb up the ladder into the ship

EXPECTED RESULT
No matter what direction you look, you should climb up the ladder

https://www.youtube.com/watch?v=JZeETUVKMoc

# ? Dec 29, 2016 11:15

Bofast: Feb 21, 2011; Grimey Drawer

Tinfoil Papercut posted:

Guys I know this is the place to talk about the greatest game ever made or to be made, but have you tried Speed Runners? It's probably even slightly better than Star Citizen, if you can believe it.

Also did they really just switch engines? loving lol, where do I pledge money for imaginary spaceships?

Speedrunners is quite good, not to mention an actual game that works.

# ? Dec 29, 2016 11:16

Bofast: Feb 21, 2011; Grimey Drawer

The Titanic posted:

As a pasty white girl who is more vampire than human due to a lack of sunlight exposure, I say it lovingly and with care. It is ok friend, I am not exactly the great outdoors' best friend here to shame anybody.

We can hardly blame you. Can't be that much sunlight down by the ocean floor.

# ? Dec 29, 2016 11:17

AP: Jul 12, 2004; One Ring to fool them all
One Ring to find them
One Ring to milk them all
and pockets fully line them; Grimey Drawer

AFTER DYING AND AWAKENING: VIEW BUGS - CHAR WALKS, VIEW STANDS

ACTUAL RESULT
Closing the complete Game and restart complete

EXPECTED RESULT
I think the server has loading probs or there a cheater on the servers.

# ? Dec 29, 2016 11:19

Mu77ley: Oct 14, 2016

Gravity_Storm posted:

'Lets sell the Millenium Falcon! Sandi can play dead Leia!'

Is her acting ability up to the level of playing a corpse though?

# ? Dec 29, 2016 11:30

AP: Jul 12, 2004; One Ring to fool them all
One Ring to find them
One Ring to milk them all
and pockets fully line them; Grimey Drawer

https://www.youtube.com/watch?v=YiFVS2EiVSc

# ? Dec 29, 2016 11:30

Tippis: Mar 21, 2008; It's yet another day in the wasteland.

Second Sun posted:

Our galaxy is only 100,000 ly across. What the hell are they trying to say?

They're trying to say that someone writing copy for their marketing and/or some PR executive who clears said copy to go live does not understand space.

I don't mean that as in the way Chris does not understand how space works (cf. the old quasar rant or his confusion over momentum and zero-g), but in the sense of not understanding the very concept of the space part of space-time. As in, they don't grasp dimensions, or the difference between distance, area, and volume. Of course, given what we know of CRobber's micromanagement tendencies, it's very possible that the latter is just a more general case of the former� That, or they think that the words �squared� or �cubed� would be too confusing for their intended audience.

Someone over at FDev calculated that, if they meant volume, they're saying that their universe has a radius that roughly reaches out to the inner portion of the asteroid belt.

Tippis fucked around with this message at 11:35 on Dec 29, 2016

# ? Dec 29, 2016 11:33

moveable shape: Oct 18, 2015

Qwertycoatl posted:

It's not exactly pointless - I think it makes sense to start off by doing that if they want to gradually merge the lumberyard features in. I can believe it being a couple of days work fighting tools and importing history, especially if their original 3.7 was some kind of packaged release that doesn't exactly match the source tree Amazon has.

The thing is though, that isn't "switching to lumberyard", that's "making the very first, easiest step in the long and difficult process of switching to lumberyard", so this can only be true if CIG are lying to exaggerate their progress

We have finally arrived at a reasonable explanation. Chris is telling the truth (like-for-like) to conceal the fact that nothing substantial happened. This is a licensing move

# ? Dec 29, 2016 11:33

it dont matter: Aug 29, 2008

Two days for two developers to copy some txt files and add the lumberyard logo to their splash screen? Seems about right for CIG.

# ? Dec 29, 2016 11:45

zcrow: May 6, 2014; Ah.. yeah... um... tup tup tup tup tup.. this is something we'll add down the line

am i too late?

# ? Dec 29, 2016 11:52

Runa: Feb 13, 2011

zcrow posted:

am i too late?

of course not

# ? Dec 29, 2016 11:58

Vincent Van Goatse: Nov 8, 2006; Enjoy every sandwich.; Smellrose

Second Sun posted:

Our galaxy is only 100,000 ly across. What the hell are they trying to say?

GIVE CRISP RIBBITS MONEY

# ? Dec 29, 2016 11:59

Tippis: Mar 21, 2008; It's yet another day in the wasteland.

zcrow posted:

am i too late?

Suggested addition: a list of names before Crytek, all crossed out � like AMD and E3 and (possibly) Intel and Kevin Costner.

# ? Dec 29, 2016 12:03

Mr Fronts: Jan 31, 2016; Yo! The Mafia supports you. But don't tell no one. Spread the word.

D_Smart posted:

Also, if you look at this bullshit chart, there is no accounting for the "50%" modification to their version of CE 3.x. Like it never even existed.

Please, allow me.

# ? Dec 29, 2016 12:04

intardnation: Feb 18, 2016; I'm going to space!

Quavers posted:

https://www.twitch.tv/dangheesling

~1500 viewers. Did have 4500 when another streamer started hosting his channel, then the viewcount started bleeding quickly. Only 280 viewers in his chat channel, so most must not have been impressed enough to move over.

e: tax

maru!

# ? Dec 29, 2016 12:16

Young Freud: Nov 26, 2006

Mr Fronts posted:

Please, allow me.

They're really should be another branch that looks like that Illfonic decision tree, but for the in-game store, because it's clearly the least buggy of any of Star Citizen builds.

# ? Dec 29, 2016 12:18

AP: Jul 12, 2004; One Ring to fool them all
One Ring to find them
One Ring to milk them all
and pockets fully line them; Grimey Drawer

Mr Fronts posted:

Please, allow me.

I like this.

# ? Dec 29, 2016 12:49

intardnation: Feb 18, 2016; I'm going to space!

Beer4TheBeerGod posted:

Isn't that an alt for Al Bester or whatever his name was? Some kind of B5 weeaboo equivalent?

I remember that melt down. Good times indeed!

# ? Dec 29, 2016 13:03

intardnation: Feb 18, 2016; I'm going to space!

Mr Fronts posted:

Please, allow me.

never give wavy hands a marker ever.

# ? Dec 29, 2016 13:42

Harold Stassen: Jan 24, 2016

Beer4TheBeerGod posted:

I guess that's what happens when you don't finish your physics ~~degree~~engine.

# ? Dec 29, 2016 13:57

D_Smart: May 11, 2010; by FactsAreUseless; College Slice

Sabreseven posted:

Just to add something, Lumberyard does not mention 64 bit precision anywhere in it's documentation and I can find no mentions of it at all. Might just be that I can't find it or that it is burried, but I'd find it hard to believe Amazon devs would have had any prior use for it.

I don't know what that means, if anything, as I'm still on holiday and quite drunken.

They don't. I checked. It's in my blog.

quote:

In addition to that, let�s play devil�s advocate again. If they have modified CE3 by 50%, and Star Engine modifications are on parity with Lumberyard to the extent that they could still keep all their modifications (e.g. Lumberyard currently has no implementation of 64-Bit positioning), then why do they need to switch? They don�t need Lumberyard in order to switch from Compute to EC2.

# ? Dec 29, 2016 14:10

Harold Stassen: Jan 24, 2016

When pieces of the truth come out like the engine swap (either they didn't modify CE or this change-over was a lot more costly in time and money than they say, I lean toward the former), I'm reminded that dreams aside, anyone who thinks Chris Roberts is an honest man deserves to have their money stolen.

# ? Dec 29, 2016 14:12

Fat Shat Sings: Jan 24, 2016

no_recall posted:

105,700,083 light years?

They made the biggest gameworld yet!

Buy an Idris!

It's hard to know what that even means since there is no giant asterisk next to it. We can only try and guess.

In my mind it's obviously a cubic kilometer figure. They are simulating a giant block of empty space, so they take the amount of kilometers in the cubic area of the space they are simulating and come up with the total kilometers to explore.

The only way I see them getting anywhere close to 100 sextillion is if they are counting a future world where they have all 100 systems finished, and multiply the amount of cubic km in the one system they do have by 100. Even then they might be using a future size figure they haven't even implemented yet.

I hate math though but that has to be it. They are counting all 100 systems and then counting each individual kilometer in whatever cubic size each individual system has based on the single system they have.

If this is true then they are just outright lying, again. It's impossible for them to multiply the amount of space they have by 100 if it's already a buggy impossible mess that barely runs and 99.999999999999999999999999999999999999999% of that space will be completely empty extending in all directions. They probably even figured out how many planets they would have if the game came out and added their cubic kilometer to the figure counting the space twice. The planet might be counted in the space it sits in but then the surface area x (amount of planets) is being thrown in there too.

Again, no giant asterisk saying "This claim is based on ______" or "at some future date" or whatever, since even the paying customers of CIG have never cared enough to verify any of their claims and they just burst into tears of pure adulation congratulating CIG on everything it claims. It's just a number someone pulled out of their rear end but my guess is it's just however many cubic kilometers it takes to involve 1 sextillion kilometers multiplied by 100 for the future amount of planned systems.

Let's see what wolfram alpha says
https://www.wolframalpha.com/input/?i=100+sextillion+kilometers

quote:

Unit conversions:
6.214�10^22 miles
1�10^26 meters
10.57 billion ly (light years)

quote:

Comparison as radius:
≈ ( 0.17 ≈ 1/6 ) � present radius of the Universe measured as a comoving distance (≈ 5.87�10^26 m )
Comparison as diameter:
≈ ( 0.11 ≈ 1/9 ) � diameter of the observable universe (≈ 93 billion ly )

Simulating 1/6th the entire universe seems like a reasonable claim for the alpha 2.6 and in no way is a scam tactic to sell spaceships. No sir.

# ? Dec 29, 2016 14:13

D_Smart: May 11, 2010; by FactsAreUseless; College Slice

ewe2 posted:

Sorry in advance for this rant, I just wanted to make it clear as clear why there are problems with CIGs/Parrys "explanation". Please point out mistakes or confusions/omissions.

Just managing the pieces of a website in git can teach you a lot about how VCS actually works and not how Ben Parry wants you to believe, like just skipping main development line history. Good VCS patterns are about minimising merge conflict and small, manageable deltas, regardless of the size of the project.

Let me break it down into a hypothetical, the kind of hypothetical that the Reddit crowd hates because they'd have to apply braincells to it, and let's pretend we're using git because its an excellent distributed VCS that lets you decide how to manage your workflow:

You've got engine code, it's made up of different bits that different groups of developers can work on, and like a good game company you want to make modifications to make the greatest game ever and you want builds to work so you can demonstrate how things are going (you know, not like SC). There will be a bunch of stuff that doesn't use the engine at all, ignore that in this discussion, it would be added at the build branch stage to make a build work and then get added to the main branch. Here we're only concerned with bits that need the engine code.

Ideally we'll have a development branch, a build branch and a main branch. You're the guy coordinating this because your developers are all working on their bits of the engine and when they put in changes, you make sure all the different bits agree and you take a snapshot of the current development branch and call that the build branch (and add those bits I mentioned above), and if it builds, those changes make it into the main branch. Clear so far? You can give these snapshots whatever name/version you like as long as it makes sense to everyone.

Ok, so down the track your builds are going fine but meanwhile the engine developer has released a new version of the engine and you want to take advantage. So you grab the new engine code, and to be safe, you clone a new copy of the original engine code and you do a merge to see what the changes are and make a diff from that. You now literally have what changed between engine versions. Well, like you would expect with a new engine version, they've changed the API you use to access its code and they've restructured its different bits quite a bit, there are new bits, older bits have gone. You do the same technique as before, comparing your current main branch (NOT the development branch, guess why) to the code and oh dear there'll be merge conflicts everywhere, the engine is quite different. What to do?

At this point we are up to BEFORE the LY fork, at 3.8 whatever version. We can break the task down: some bits can be easily discarded, some bits will be added, but there'll be a chunk that will have to be integrated piece by piece preferably along structural lines and meanwhile you have to educate the developers who can't continue along the lines of continuous development/builds until these changes are made. You have to educate the developers on the changes to the API, you have to reassign developers to the bits that are changed/new, and you may have to take a deep breath and throw a lot of good code away simply because you don't have the time or resources to rebase everything. "Rebase" sounds simple but only works if your modifications are only in the parts of the new code that didn't change, and in the case of an engine only some of that will be true.

Ideally, you can give some developers the freedom to rebase their own development work on the new engine code, but in some cases you'll have to do the work yourself and get them to clone from that. And THEN it's back to making the builds work, having "rebased" on the new code and then you've got a main branch which is a fork off the engine code.

And THEN you have the LY fork. Take the above scenario and repeat, unless you're prepared to simply throw away all the previous work. Unless the LY fork is amazingly compatible which I do not believe for a second.

I want to point out that as far as I understand, game development is rarely this organized or as careful. An object hierarchy is fine for business software development but usually only gets as far as the UI and maybe file operations but everything else is written pedal to the metal good old functional programming and that is a dog to manage even with the help of a VCS. The whole point of object orientation was to make large projects even feasible but that breaks down with the way games work which do multiple things at the same time like a mini-OS, it's that complex.

So when Ben Parry tries to tell me that it was easy I'm just dumbfounded. Because he should know better. And this discussion is only applicable to if they did it as he claims, ripptide might be closer to the truth.

# ? Dec 29, 2016 14:15

Syd Syko: Nov 15, 2012; Out there ----> Somewhere

Tokamak posted:

For a standard licensing arrangement? Why would a startup studio with a few million dollars (at the time) pay more for a multi-game/unlimited license when they were originally planning to ship one game? They may have since secured the appropriate licensing, but businesses have a tendency to try and stiff their supplier, and hold out paying their invoices until the last possible moment. CIG has made it a point that they haven't been receiving Crytek support for years now. It doesn't paint a good relationship between CIG and Crytek.

Where is the source that they own the engine or shelled out for a more comprehensive license? This presupposes that they knew in advance that they would be making the $20+ million expanded scope Star Citizen, and three single player games. The original scope was designed to be built in the basic CryEngine with limited modifications.

Ok, so what you're saying is, you don't actually know poo poo about what they did or didn't do, and are basing your bullshit off of some poo poo you read on the internet. Uh, ok, whatever dude, toodles~

# ? Dec 29, 2016 14:20

D_Smart: May 11, 2010; by FactsAreUseless; College Slice

Mr Fronts posted:

Please, allow me.

Baxta posted:

This is a good post and explains exactly why the only thing they could have possibly done is used LY for the 3.7 fork making the whole transition absolutely pointless (Unless they are stupid enough to think LY will somehow allow them to produce their insane promises)

Though that still will not account for the 50% modification to the engine they claimed to have done. :allears:

# ? Dec 29, 2016 14:22

Samizdata: May 14, 2007

ewe2 posted:

Sorry in advance for this rant, I just wanted to make it clear as clear why there are problems with CIGs/Parrys "explanation". Please point out mistakes or confusions/omissions.

Just managing the pieces of a website in git can teach you a lot about how VCS actually works and not how Ben Parry wants you to believe, like just skipping main development line history. Good VCS patterns are about minimising merge conflict and small, manageable deltas, regardless of the size of the project.

Let me break it down into a hypothetical, the kind of hypothetical that the Reddit crowd hates because they'd have to apply braincells to it, and let's pretend we're using git because its an excellent distributed VCS that lets you decide how to manage your workflow:

You've got engine code, it's made up of different bits that different groups of developers can work on, and like a good game company you want to make modifications to make the greatest game ever and you want builds to work so you can demonstrate how things are going (you know, not like SC). There will be a bunch of stuff that doesn't use the engine at all, ignore that in this discussion, it would be added at the build branch stage to make a build work and then get added to the main branch. Here we're only concerned with bits that need the engine code.

Ideally we'll have a development branch, a build branch and a main branch. You're the guy coordinating this because your developers are all working on their bits of the engine and when they put in changes, you make sure all the different bits agree and you take a snapshot of the current development branch and call that the build branch (and add those bits I mentioned above), and if it builds, those changes make it into the main branch. Clear so far? You can give these snapshots whatever name/version you like as long as it makes sense to everyone.

Ok, so down the track your builds are going fine but meanwhile the engine developer has released a new version of the engine and you want to take advantage. So you grab the new engine code, and to be safe, you clone a new copy of the original engine code and you do a merge to see what the changes are and make a diff from that. You now literally have what changed between engine versions. Well, like you would expect with a new engine version, they've changed the API you use to access its code and they've restructured its different bits quite a bit, there are new bits, older bits have gone. You do the same technique as before, comparing your current main branch (NOT the development branch, guess why) to the code and oh dear there'll be merge conflicts everywhere, the engine is quite different. What to do?

At this point we are up to BEFORE the LY fork, at 3.8 whatever version. We can break the task down: some bits can be easily discarded, some bits will be added, but there'll be a chunk that will have to be integrated piece by piece preferably along structural lines and meanwhile you have to educate the developers who can't continue along the lines of continuous development/builds until these changes are made. You have to educate the developers on the changes to the API, you have to reassign developers to the bits that are changed/new, and you may have to take a deep breath and throw a lot of good code away simply because you don't have the time or resources to rebase everything. "Rebase" sounds simple but only works if your modifications are only in the parts of the new code that didn't change, and in the case of an engine only some of that will be true.

Ideally, you can give some developers the freedom to rebase their own development work on the new engine code, but in some cases you'll have to do the work yourself and get them to clone from that. And THEN it's back to making the builds work, having "rebased" on the new code and then you've got a main branch which is a fork off the engine code.

And THEN you have the LY fork. Take the above scenario and repeat, unless you're prepared to simply throw away all the previous work. Unless the LY fork is amazingly compatible which I do not believe for a second.

I want to point out that as far as I understand, game development is rarely this organized or as careful. An object hierarchy is fine for business software development but usually only gets as far as the UI and maybe file operations but everything else is written pedal to the metal good old functional programming and that is a dog to manage even with the help of a VCS. The whole point of object orientation was to make large projects even feasible but that breaks down with the way games work which do multiple things at the same time like a mini-OS, it's that complex.

So when Ben Parry tries to tell me that it was easy I'm just dumbfounded. Because he should know better. And this discussion is only applicable to if they did it as he claims, ripptide might be closer to the truth.

You are making EXACTLY my point, albeit with MUCH more effort involved. The Citizenry is deluded they know anything about multi-threaded development because of the endless rounds of buzzword bingo with Chris and Company, so they happily accept any vaguely technical sounding explanation as long as it is, at some point, explained with adequate flailing.

# ? Dec 29, 2016 14:25

Adbot: ADBOT LOVES YOU

# ? May 9, 2024 10:58

Samizdata: May 14, 2007

DancingShade posted:

lol as if I'd lower myself to make a reddit account.

Unlike some people here.

tooterfish posted:

Stop loving swearing.

Hey, now, some of us slog through the trenches, knee deep in blood and poo poo, because it needs be done. Warfare is not effective waged purely from deep behind friendly lines. Am I right, Derek?

# ? Dec 29, 2016 14:30

The Something Awful Forums > Archives > Comedy Goldmine > The Goodmine > Star Citizen: Death of a Space Thread?

«‹›6530 »