* eliminate some more transmute()
* Address `review comments`
* Make unsafe block smaller
* Use different byte-order than `std`, read words as big endian instead
of little endian!
* Fix IpAddresses nits
* Use `from_be` to work both for big and little endian
* Ipv6 addresses were incorrectly `transmuted`
* remove needless lifetime annotation
* Update `add_license` script
* run script
* add `remove duplicate lines script` and run it
* Revert changes `English spaces`
* strip whitespaces
* Revert `GPL` in files with `apache/mit license`
* don't append `gpl license` in files with other lic
* Don't append `gpl header` in files with other lic.
* re-ran script
* include c and cpp files too
* remove duplicate header
* rebase nit
* Rebase and fix compilation errors (tests not yet fixed)
* Use `debug_assert` over `assert`
* Fix tests
* Assert safety, clean up
* Fix up stale cache removal, move one assert to debug_assert
* Remove printlns
* Add licenses
* Fix benches
* Inline some no-ops in a hot loop that weren't being inlined
* Add spooky comment to make sure no-one removes the inlining annotations
* Minor cleanup
* Add option to switch between mmap and ram
* Flag ethash to use less memory when running light client
* Fix tests
* Remove todo comment (it's done)
* Replace assertion with error return
* Fix indentation
* Use union instead of `transmute`
* Fix benches
* Extract to constants
* Clean up and fix soundness holes
* Fix formatting
* Ignore missing-file errors
* Make incorrect cache size an error condition instead of a panic, remove dead code
* Fix compilation errors from rebase
* Fix compilation errors in tests
* Fix compilation errors in tests
* Fix benchmarks
* Fix unsoundness in uses of unsafety
* Remove most uses of unsafe indexing and ptr::copy_nonoverlapping
This commit also includes a completely absurd optimisation that I
promise is an honest win. You can check the benchmarks, I barely
believe it myself.
* Add safety comment
* Add more safety comments
I also tried unrolling the 256-iteration loop further below, but it actually caused a
slowdown (my guess is either branch prediction stopped kicking in or the instruction
cache was being maculated).
Implementations of get_cache_size and get_data_size in Rust (issue #161)
Removed sizes module, containing replaced data tables
Fixed whitespace issues after code review
