Binary, List and Tuple Inequalities (Paradox?)

[Fred Hebert]

On Fri, Oct 25, 2019 at 11:35 AM Dmitry Belyaev <be.dmitry@REDACTED> wrote:

> To me alphabetical ordering of lists (as legacy string representation) and
> binaries (as modern string representation) is natural, so it is expected
> that "aa" < "b" < "bb".
>
> As a general note, this is only true to the extent that you don't care
that "ZOOM" is < than "apple" in this dictionary. It is anything but
natural, I would say it's probably more "accidental". Actual string sorting
tends to rely on a Unicode collation algorithm (a part of the spec that is
still missing in the stdlib), since even string normalization won't be
enough to ensure things are done safely with regards to human expectations.
At this point only libraries such as ux
<https://github.com/erlang-unicode/ux> support it well.

Most of the orderings we have right now are often arbitrary and can have
surprising sides to them. At this point for most data structures the
important thing is not that the ordering is correct, but that at least one
exists at all; this allows the creation of a bunch of data structures
(usually trees and variants) to work on all data types, mixed or not. All
atoms are smaller than all strings, even if atoms have a string component
to them, for example.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20191025/5baa3573/attachment.htm>