There are some hacks to allow toHash() methods
to be const nothrow @safe, these are due to
Phobos problems (e.g. Variant.toHash() is not const nor nothrow).
After experiments with loading the whole file at once, and
with decoding and parsing in separate thread, lazy reader
turned to be the fastest/least memory intensive solution.
Characters are now decoded in small batches.
This improved parsing speed by ~20%.
No global state anymore. Anchors are now zero terminated strings
and TagDirectives are a simple array. Event structure was changed
to prevent size increase.
Minor fixes and improvements.
slowdown and a slight memory usage increase. Fixed a potential
bug in Reader. Tag is now implemented as a zero-terminated
string, removing the need for global state.
Node opIndex now returns a reference.
as loading. Also slightly improved loading performance.
Greatly improved performance with very small files. We can now
load 10000 24 byte files in a second, and dump them at the same
speed.
Fixed another FastCharSearch bug.
ordinary array, added a mixin to generate fast lookup tables for
character search, and used said mixin in the greatest
bottlenecks. Another great speedup, at least 20%, can't keep
track of it anymore.
Removed unnecessary calls to Stream.available(),
again for grat speed gain. Also various small optimizations.
Overall, improved loading speed about 400%.
