IAHM-COL wrote in Thu May 07, 2015 8:49 pm:The complete FGDATA next with submodules is currently 33G (with the .git folder, so the real git repository size). The "./Aircraft" directory (which is FGAddon after all additional aircrafts and livery expansions) is 14G.
Still waiting for the key number -- which is the increase in size due to the livery expansions. Can you give us a before and after value please?
IAHM-COL wrote in Thu May 07, 2015 8:49 pm:3. We can come back to speak about isolating the liveries in a branch, as you suggest. But importing them is the first step required to open that conversation, and have some real numbers to discuss. It may be unnecessary to open a new branch for most aircraft with 1 or 2 additional liveries only, if you see where I am going.
I commend your efforts and I agree that doing something is better than just talking about it... but sometimes planning and having a healthy discussion (as we're currently having) is valuable before you implement a feature which takes a lot of time. Why? Because once you've done the work you become more emotionally invested in your idea, naturally -- you don't want to give up on work you've already done.
IAHM-COL wrote in Thu May 07, 2015 8:49 pm:4. Create the branch, adds an additional problem. The branch doesn't follow the development tip automatically
Correct. Every major (or minor, if you so please) release you do two commands:
git checkout all-liveries
git pull origin master
not so big of a problem, really.
IAHM-COL wrote in Thu May 07, 2015 8:49 pm:5.The solution I had everytime spoke about is modularity. Really, you will just install (init+update) those aircraft that fit your interest. So that will solve most of your repository size problem.
What's the point of having everyone's livery if I can't have everyone's plane? It defeats the entire purpose.
IAHM-COL wrote in Thu May 07, 2015 8:49 pm:6. Isolating the liveries on its own branch has a very neglectible effect on the "repository" size. Actually. Because the repository, if you do a complete clone (as oppose to a single-branch approach) will contain ALL of the objects. So we have the same "net" size for No-liveries. ==Think about this point and if this point is unclear, lets talk about it again, since it is critical here.
Fair point except most end-users don't use git clone, they use the download zip button, which only downloads the branch you're on (master, by default). Even I, as a developer mostly use download zip for the exact reason you state -- I do not need (or want) 5 versions of the DC3 or DHC6 (which is also why I have suggested seperate repo for separate versions of the plane... but that's another discussions so let's not bring that up).
IAHM-COL wrote in Thu May 07, 2015 8:49 pm:I also want you to take a look at your values. And they definitely don't correspond to triple or more of the aircraft size (which Gijs indicated do occur for a few exceptional cases , like the dhc6)
also, keep in mind that your numbers are telling us what is the size of the livery package without the extra liveries. What would be the size difference* after the extra liveries are installed, is what comes to more critical here.
*size of Models directory after extra liveries installed - Size of Models directory before extra liveries are installed
My point, which I guess wasn't clear enough is that even WITHOUT the extra liveries, the default ones ALREADY take up a large amount of the download size. This is probably why most aircraft developers in the past haven't included all liveries in the past 19 year or however many years FG's been around.
Give me a few minutes, and I'll give you the values WITH the extra liveries.
Best,
Phil
P.S. This is a very constructive discussion. I hope we can investigate the numbers, and evaluate whether this idea (which in theory is great) will provide value to the community, or just cause greater file sizes (which for some people is a problem).