Bugfix release 2.1.6

7 views
Skip to first unread message

Sergey Basalaev

unread,
Oct 20, 2013, 9:45:02 AM10/20/13
to Alchemy OS
This release fixes bugs which are quite serious and it remains mystery for me why they were unnoticed for so long.
Bugs fixed in this release:
* Hexadecimal character literals (like '\u03C0') were parsed incorrectly by compiler
* asin() and atan() functions returned incorrect values

Since GoogleCode stops supporting new downloads, Alchemy OS can now be downloaded from SourceForge.

Kyle Alexander Buan

unread,
Oct 20, 2013, 9:49:46 AM10/20/13
to alche...@googlegroups.com
Thanks, Sergey!
Just wondering, though. How did you notice the problem with atan() and
asin()? Doing some trigonometry? :)
> --
> You received this message because you are subscribed to the Google Groups
> "Alchemy OS discussion group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to alchemy-os+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>

Sergey Basalaev

unread,
Oct 20, 2013, 11:06:09 AM10/20/13
to Alchemy OS

2013/10/20 Kyle Alexander Buan <tar.s...@gmail.com>

Thanks, Sergey!
Just wondering, though. How did you notice the problem with atan() and
asin()? Doing some trigonometry? :)

While refashioning core libraries for the next major release I've decided to
test how accurate my implementations of math functions are and found
out that the maximum difference between my atan() and Math.atan() is
about 0.26 - VEEERY inaccurate. asin() and acos() are both computed
using atan() so they are also affected.

Kyle Alexander Buan

unread,
Oct 20, 2013, 11:49:30 AM10/20/13
to alche...@googlegroups.com
Great! When it comes to math, every degree of accuracy counts ;)

On 10/20/13, Sergey Basalaev <sbas...@gmail.com> wrote:

Sergey Basalaev

unread,
Oct 21, 2013, 1:59:57 AM10/21/13
to Alchemy OS
2013/10/20 Kyle Alexander Buan <tar.s...@gmail.com>
Great! When it comes to math, every degree of accuracy counts ;)

Yeah, usually error in the first digit is not what you expect.
Reply all
Reply to author
Forward
0 new messages