(much) faster to read the decompressed files as opposed to reading them
in-place from the jar. At least I double-checked before going too
far. In case you were wondering, it's *really* fast to simply check
a jar file for an entry that doesn't exist.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1165 ed5b42cb-e716-0410-a449-f6a68f950b19
near as I can tell, it's substantially faster to read resources
directly from the jar file rather than unpacking them first. Why
would that be? Well, for starters, I'm guessing it takes less time to
decompress data than to read more data from the storage device. Also,
it's likely to use disk space more efficiently: having all the
resources in one jar makes it more likely that they'll occupy adjacent
blocks, not to mention the fact that there's less wasted space from
partially filled blocks. So, add an option to avoid unpacking files
from jars before reading them. This doesn't work with FastImageIO's
memory-mapped image reading, but we don't use that in Spiral Knights.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1164 ed5b42cb-e716-0410-a449-f6a68f950b19
pending, (ie. set to play but hasn't finished buffering)
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1160 ed5b42cb-e716-0410-a449-f6a68f950b19
include the model/animation exporters for 3DS Max.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1149 ed5b42cb-e716-0410-a449-f6a68f950b19
an identity hash set to accelerate it (identity hash set uses linear probing,
which can give better cache performance than the chaining used by the normal
hash set).
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1129 ed5b42cb-e716-0410-a449-f6a68f950b19
tested and debugged. Turns out the problem I was chasing was a digester-2.0
dependency where a digester-1.8 dependency was needed, but I'll go ahead and
commit this anyway.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1123 ed5b42cb-e716-0410-a449-f6a68f950b19
viewer, so now we have colorized pirates walking around the scene instead of
weird red/purple ones.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1060 ed5b42cb-e716-0410-a449-f6a68f950b19
up will be making the tests actually test something, and then fixing that when
they break.)
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1050 ed5b42cb-e716-0410-a449-f6a68f950b19
getting its image and doing some math, which fucks us if we're working with
bogus tilesets (which we do when testing).
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1049 ed5b42cb-e716-0410-a449-f6a68f950b19
forthcoming to fix up the tests.
Peskily, it looks like I'm going to have to take one for the team and become
the maintainer of a Maven artifact for LWJGL. In spite of repeated requests
over the years for Mavenized artifacts for LWJGL, Mazon resists it (he seems to
think he has to use Maven to *build* LWJGL, which he certainly need not do).
There have been a variety of half-assed attempts to maintain third-party Maven
repositories, all of which have petered out a major version or two ago.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1045 ed5b42cb-e716-0410-a449-f6a68f950b19
src/main/java. Prepared to move tests out into main project.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1040 ed5b42cb-e716-0410-a449-f6a68f950b19
code into the Bang project directly so as to confine the dependencies on JME
libraries to the only thing that will ever need them.
git-svn-id: svn+ssh://src.earth.threerings.net/nenya/trunk@1035 ed5b42cb-e716-0410-a449-f6a68f950b19