Apparently my -O3 came from somewhere other than this script.
It's possible that there is some generic "use optimisations"
mechanism in setuptools/distutils/whateverthisis, but I don't feel
like opening up that can of worms so let's just add an -O3 extra
argument to the compiler invocation to make sure it always gets it.
Older versions apparently didn't have LANCZOS filtering for
resizing. In this case, just fall back to whatever the default is.
Will look marginally worse for those people but at least it works.
If one wishes to add debug output, they should use overviewer.utils.debug.
To see the output produced by that, set overviewerConfig.map.debug to true.
Fixes#1816.
Previously, walls were all numbered at 21000+. This is bad because
our blockmap is an array, so this caused our blockmap to take up
80 MiB of RAM in each worker process.
This commit changes the wall numbering, and exploits some bitmasking
to have the wall check run in constant time no matter how many walls
there are. This is done with a simple mask and xor to check the prefix.
RAM usage for the blockmap thus drops to like 44 MiB.
This is in preparation for adding more walls for 1.16.
Instead of trusting that Minecraft will only request up to index
num_palette_entries - 1, we'll assume Mojang occasionally produces
a weird file where it tries to request past that, up to the maximum
palette index it can based on the bits per value. To facilitate this,
this commit extends the translation palette array to this maximum size.
Such invalid palette picks will simply be replaced by ID 0 (aka air),
so it should cause minimal weirdness in the output.
Works around #1797.
Minecraft 1.16.1 loves writing these in singleplayer, and they
cause a whole bunch of corruption warnings if we don't handle them
in a special way like this.
using rem here is probably wrong because we want to be the height
of the line in the span we're inside of, not of some global thing.
Also, add a margin to the left.
After discussing this with the other folks on IRC, everyone seemed
to be cool with me adding myself to this. It makes it so people see
a little sponsor button that tells them how they can monetarily
support the project if they choose to do so.
Minecraft occasionally generates chunks which are not yet lit.
In the past, I'd have said to just make them not render, but these
can sometimes be large areas of the world.
Instead, render them with full bright SkyLight. This looks less bad
than whatever is stored in the SkyLight property in these cases.
Closes#1787, probably. Only one person bothered providing a sample file.
Closes#1790.
If a chunk fails reading due to an underlying operating system
IO error, we can invoke the retry logic too. No clue what could
cause this to happen beyond hardware faults.
Also lower the sleep time to 0.25s down from 0.5s because half a
second of sleeping for each chunk this happens on seems incredibly
overkill.
Making the nbt code raise a CorruptChunkError in this case also means
that failing to read a chunk due to an IOError is not a fatal error
anymore, it'll simply skip it and move on.
Sometimes people's texture packs contain weird things, and it'd
be good if instead of throwing an unhandled IOError, we raise
a TextureException with the filename in it.
Also, make verbose mode actually run find_file in verbose mode,
which is useful for finding where Overviewer loaded that texture
file from.
This reverts commit 9895fe875b.
The original code was correct, but confusing. Overviewer did regenerate
the texture in each worker process because Python pickling is incredibly
fragile fucking garbage designed by idiots who have no clue of what
a good language looks like.
We used to generate textures multiple times for each worker process,
which didn't seem to be intended as it was done gated behind a reversed
if statement. It just so happened to fix an issue introduced by the
pickling code which was deleting vital attributes like the actual
generated textures.
I'm honestly not sure why this was ever done, so let's just throw this
out and call it a day.
Fixes#1728
People could always set this through the config file, but somehow
not the command line. This commit fixes this oddity by adding the
option to the genPOI script's argument parser.
Don't use .append(), we already know the final length. Instead,
have python initialise the list of a certain size and set the elements
in it.
Makes the function like 4% faster.
Turns out my previous grasp of how Minecraft does this was wrong.
This seems to be the correct way. One side effect is that biome data
now has less resolution. One only really notices this when looking at
water, for which Minecraft does not even use the water colours for
in-game, otherwise I can't really tell a big difference.
Fixes#1698.
Okay so you know how you have chunks and they map to 4 levels of biomes?
Logically you'd go hmmmm, let's just map 4 Y levels to one biome level.
Except there aren't 16 chunk heights, there are about 17 if you count
Y=16 (which can contain data), but there's also -1 technically but I've
never seen this have data in an NBT structure I don't think.
Anyway, let's just hope we're doing the right thing by giving Y=16 the
same biome as Y=15. If someone wants to check whether that is actually
correct please feel free to, but for now it's better than crashing at
the very least.
Would be nice if we had a spec for these kinds of things.
Fixes#1685.
Apparently rotated worlds would try to access the biome for Y-level
-1. I don't know why they do that (probably some dumb reason that may
be a bug in of itself) but returning the first level should be harmless.
Fixes#1683.
Fixes#1650.
Mojang changed the biomes code so that it now can have different
biomes for different Y levels. We need to adjust our logic accordingly,
which is done through some small BiomeDispensary class where we shove
a numpy'd Mojang array in and can then read out the biomes for each level.
Biome data is now stored per-section, which needed some changes on the C
side of things. I didn't change anything in the biome overlay code so
I wouldn't be surprised if it's broken now, but for the time being I'd
rather have 1.15 fixed than some obscure overlay.
Tested to work with 1.14 and 1.15 data. No new biomes have been added
to the code yet.
Observers get initialised at config parse time, before Overviewer
creates the outputdir. So we can just tell them to create the
outputdir if it doesn't already exist to avoid an awkward error
on the initial render.
Fixes#1657.
Nicolas F