[Caml-list] [ANN] Uucp 0.9.0

[Caml-list] [ANN] Uucp 0.9.0

[Daniel Bünzli]

Hello, 

I'd like to announce uucp:

  Uucp is an OCaml library providing efficient access to a selection
  of character properties of the [Unicode character database][1].

  Uucp is independent from any Unicode text data structure and has no
  dependencies. It is distributed under the BSD3 license.

  [1]: http://www.unicode.org/reports/tr44/

Home page: http://erratique.ch/software/uucp
Documentation: http://erratique.ch/software/uucp/doc/Uucp
Github mirror: https://github.com/dbuenzli/uucp

It should be available shortly in opam. A few comments about the
library can be found at the end of this message.

Part of this work was sponsored by OCaml Labs, many thanks for their
support.

Best,

Daniel


A few notes.

* Having seen unwarranted sights of horror at the simple mention of 
  Unicode by fellow peers I took the time to write an absolute minimal
  introduction to Unicode in the documentation:

  http://erratique.ch/software/uucp/doc/Uucp.html#uminimal
  
  At the end of this introduction end I also give a few biased tips on
  how Unicode can be handled in OCaml.

* The data used to represent the properties is directly linked
  in your executables. In the future I will adapt the library to use
  OCaml 4.02 module aliases so that you only pay for the submodules
  you access. For now linking against the library on osx 64-bit, 
  result in a 6.4 Mo executable and on linux 32-bit a 3.7 Mo
  executable.

* If you are interested in Unicode caseless matching (equality) 
  of strings or identifiers, the documentation of the Uucp.Case module
  has sample code on how to perform that using Uutf and Uunf. This
  code may eventually be gathered in a proper module in the future.

  http://erratique.ch/software/uucp/doc/Uucp.Case.html#caseexamples

* Not all properties are exposed. Obsolete properties, deprecated
  properties and those that are specific to some unicode processing
  algorithm (bidi, segmentation, normalization, etc.) are left out.
  The reason for the latter is that these algorithm may need to devise
  their own, maybe more efficient, representations. For example the
  normalization properties are not included, as they are best stored,
  used and exposed by a module that performs normalization
  (e.g. Uunf). It is not excluded that I reverse this in the future
  and make Uunf dependent on Uucp as this could make maintenance
  easier (though Uunf sometimes uses optimized representations). 
  Also if the data needs to be used by more than one module
  this may become less wasteful than each module including its own
  data.

* Regarding the last point, the selection is subjective. If there's 
  a property you feel is useful in the wide and missing please 
  tell me (e.g. on github's issue tracker) with a good rationale and I 
  will add it. The full list of omitted properties is available here:
  
  http://erratique.ch/software/uucp/doc/Uucp.html#distrib_omit

Re: [Caml-list] [ANN] Uucp 0.9.0

[Goswin von Brederlow]

On Sat, Jun 28, 2014 at 11:08:00PM +0100, Daniel Bünzli wrote:
> Hello, 
> 
> I'd like to announce uucp:

http://en.wikipedia.org/wiki/UUCP

Maybe not the best name.

MfG
	Goswin

Re: [Caml-list] [ANN] Uucp 0.9.0

[Daniel Bünzli]

Le lundi, 30 juin 2014 à 09:25, Goswin von Brederlow a écrit :
> http://en.wikipedia.org/wiki/UUCP
>  
> Maybe not the best name.
I think you are a little bit short sighted.

1) The UUCP protocol suite is now mostly a historical artefact

2) The library is called uucp and the module Uucp, not UUCP, just don’t apply Unicode’s canonical caseless matching algorithm and you will be able to distinguish...

3) Uucp as a name perfectly fits among the names of my modular implementation of Unicode (Uucd, Uutf, Uunf and more in the future)

4) Given the structure of the library and the way it’s going to be used in code, I doubt it will cause any confusion among people reading code that uses the library (recurse to point 1)).

Best,

Daniel

Re: [Caml-list] [ANN] Uucp 0.9.0

[Goswin von Brederlow]

On Mon, Jun 30, 2014 at 10:37:43AM +0100, Daniel Bünzli wrote:
> Le lundi, 30 juin 2014 à 09:25, Goswin von Brederlow a écrit :
> > http://en.wikipedia.org/wiki/UUCP
> >  
> > Maybe not the best name.
> I think you are a little bit short sighted.
> 
> 1) The UUCP protocol suite is now mostly a historical artefact

It's still being used though.
 
> 2) The library is called uucp and the module Uucp, not UUCP, just don???t apply Unicode???s canonical caseless matching algorithm and you will be able to distinguish...

Package: uucp
Description: Unix to Unix Copy Program

Package: ocaml-uucp
Description: Unix to Unix Copy Program for OCaml (runtime)

So what (Debian) package name should your Uucp then have?
 
> 3) Uucp as a name perfectly fits among the names of my modular implementation of Unicode (Uucd, Uutf, Uunf and more in the future)
> 
> 4) Given the structure of the library and the way it???s going to be used in code, I doubt it will cause any confusion among people reading code that uses the library (recurse to point 1)).
> 
> Best,
> 
> Daniel

I'm just saying that it will cause confusion, at least among old
timers like me.

MfG
	Goswin

Re: [Caml-list] [ANN] Uucp 0.9.0

[Daniel Bünzli]

Le lundi, 30 juin 2014 à 15:11, Goswin von Brederlow a écrit :
> Package: ocaml-uucp
> Description: Unix to Unix Copy Program for OCaml (runtime)

Can’t  find it here:

https://packages.debian.org/stable/allpackages

> So what (Debian) package name should your Uucp then have?
According to the Debian OCaml packaging policy [1] it should be libuucp-ocaml.  

> I'm just saying that it will cause confusion, at least among old
> timers like me.

Given that UUCP is no longer widespread, context should be largely sufficient to avoid confusion.

Best,

Daniel

[1] http://pkg-ocaml-maint.alioth.debian.org/ocaml_packaging_policy.html/c305.html