[Ready] Dynamic submap loading

[kevingranade]

Here's the payoff, this makes map loading incremental, which means FAAST.
Additionally saving speed is improved, since it only has to write out submaps that you've been near. Distant submaps remain untouched.
Still to be done is writing submaps out to disk and unloading them to keep memory footprint down, but that can wait for another PR.
The naming for the files is slightly changed, so this won't be compatable with the previous PR.

[dwarfkoala]

He just did it.

wow.

[KA101]

Test-compiling now.

[kevingranade]

Still a few issues, it spams map loading dialogs when you're walking around, and I've seen an occasional segfault. I think it's failing to fully populate the active grid sometimes now, probably just need to find a good place to stick a map::load().
Also I need to look at unloading some of those submaps after converting them to keep the memory pressure down a bit.

[Zireael07]

bounce bounce

[i2amroy]

Sweet! Nice to see this up and (mostly) ready. :D

[dwarfkoala]

I say that this has to be in by 0.A. The ability to actually have really really big worlds without the horrendous slowdown and eventual death of worlds is important enough, I think, that we could delay 0.A if necessary.

[KA101]

Kevin wanted it in too.

[Zireael07]

0.A is already delayed quite a lot and I agree that it will make a lot of players happy - I mean, the balance changes and the ability to play without "omg the world crashed for ever because it's too big"

[kevingranade]

This should be good to go now, but please put it through a bit of a wringer if you're able. I'm particularly interested in performance on large maps on windows.
When you load a world for the first time with this version, it will read in the whole map, write out the chunked-up map, and then delete the old maps.txt file. It does this so if it crashes when writing the save the maps.txt file will be untouched. Also it throws away the loaded submaps as it writes them to disk to try and reclaim some RAM.
In the future it will also do this if we bump the map version number, because it has to convert all the files over to the new version.

[BevapDin]

One issue: the world is not reset/deleted because the code in game::delete_world does not work recursive and only deletes files in the world folder, not in the new maps folder. But creating a new world with the same name will load the already existing maps just fine.

[KA101]

Second issue: looks like the default-mod list doesn't play well with the directory-assurance code. If you've a better way to resolve that conflict, please do.

http://pastebin.com/QzCPFbrw

[BevapDin]

@KA101: In mod_manager.cpp, line 29: the declaration of assure_dir_exist should be removed, as it is now declared in file_wrapper.h that should fix this error.
The declaration was moved and is not recognized by this branch.

[KA101]

Gave it a once-over with the megafile. Cuts save/load times well over half and retains content. teleporting around suffered no lag over that from usual teleporting.

If I don't hear a reason not to merge this by 11 PM US Eastern, it's going in.

[Poetich77]

The game stops when appears pop up message "Please wait at the map saves"

[kevingranade]

I'm guessing this is due to autosave, if not let me know.
Do you mean it stops indefinitely, or a long time?

[Poetich77]

Yes, it stops indefinitely. It happen when i'm trying to load my save game

[dwarfkoala]

I'll make an issue for that

[dwarfkoala]

Hrm, Poetich77, can you please post a copy of the savegame? For all we know it could be a problem with that specific savegame.
EDIT: https://www.dropbox.com/s/xa612d93yaxfh44/save.rar didn't see that one.

[KA101]

I'd noticed that it hung a bit, but it completed in a second or two. Chalked it up to my laptop being low-powered. Sorry, Poetich.

[dwarfkoala]

Curiously, it seems to be hanging indefinitely on my end.
I'm using XP, but my computer is horrendously underpowered so that might be
a factor?

[KA101]

XP here too. Lemme compile everything since last night and I'll take a shot at it.

[kevingranade]

What exactly is the way to reproduce this, is it just from loading a
particular save?

[dwarfkoala]

A fix is in the works.
#5948
Seems to be indeed when particular saves are loaded.

On Wed, Feb 5, 2014 at 6:30 PM, Kevin Granade notifications@github.comwrote:

What exactly is the way to reproduce this, is it just from loading a
particular save?

Reply to this email directly or view it on GitHubhttps://github.com//pull/5840#issuecomment-34286391
.

[KA101]

Yeah, that save was borked or something. Fix seems to do the job, I just merged it.

[Poetich77]

It`s working in the latest build #802! Thanks for your help, guys!