I seem to be agreeing with you re sticking to Boost licence. Only
suggesting what can be done to allow the library to do that. The concepts I
described apply in many jurisdictions (I don't presume to know all of them).
However for certitude the options are more limited.
One is to "clean room" functionally equivalent table/code given
information/primary sources needed to do so (i.e. the spec rather than any
machine usable form of it). This is potentially impractical to do and
maintain (including testing that all is correct and consistent with unicode
spec).
Another is to NOT include the unicode.org headers/data in the lib itself
but to have a compile time dependency on them, some doc's re downloading or
referencing them where (potentially) already installed on a system, and ref
to the unicode licence, which each user is free to assess for themselves.
This approach avoids the need to make the (probably invalid in most
jurisdictions) "act of download is unequivocal acceptance" licence text
part of a boost distribution and avoiding a need to build equivalent data
independently or argue about licence validity, copyright of tabulated data
necessary to conform to a spec etc.. Each Boost user can make that call
without involving Boost. For this potential user at least, that is fine.
Are there any other options? A lib that has to deliver conformance with a
spec that has licence terms should not be rejected due to its dependency on
that spec. Neither should depending on that spec cause any Boost code,
doc's etc to have a different licence. I'm out of suggestions for achieving
that.
On Sat, 20 Jun 2020, 12:37 am Damian Vicino,
El vie., 19 jun. 2020 a las 0:45, Darryl Green via Boost (< boost@lists.boost.org>) escribió:
I also am not a lawyer but know from experience that you can't copyright data, or simple expressions of it. A program is an original work. A table of facts it uses is not. Someone can write what they like in a licence but if the table can be (ideally, has been) produced manually or algorithmically from information (published in whatever form, even if copyright) that simply conveys facts you can use it. There is some risk/issue if you use the way the facts are arranged "creatively" e.g. as a graph or figure or if included in a database (that is organised "creatively" to allow use/retrieval) etc.. Also copyright does not protect an underlying algorithm - only the program that implements it. And if there is simply/only one obvious way to represent said algorithm etc it doesn't prevent one writing the same thing... Obviously a whole program composed of fragments that aren't individually copyrightable but "creatively arranged" in some original program can't then be arranged into a "new" program to form a whole that is - effectively - the original work without that copying of creative arrangement being a breach of copyright ...
The problem is that what is copyright, what is copyrightable, and what is legal or illegal is defined in a country by country basis. Exploring a WW application/interpretation of law is extremely expensive and that is why when an org settles on a license it sticks to it to keep it simple and avoid risk of overlooking something in the process and get in big trouble somewhere.