[pyicu-dev] towards release 1.0

[Andi Vajda]

On Apr 1, 2010, at 0:47, F Wolff <friedel at translate.org.za> wrote:

> Op Wo, 2010-03-31 om 15:19 -0700 skryf Andi Vajda:
>> On Mar 31, 2010, at 14:49, F Wolff <friedel at translate.org.za> wrote:
>>
>>> Op Wo, 2010-03-31 om 10:29 -0700 skryf Andi Vajda:
>>>> On Wed, 31 Mar 2010, F Wolff wrote:
>
>>> The first thing that strikes me, is that I'll need to get familiar
>>> with
>>> the C API before I can use this.
>>> Would it be possible to get some of the
>>> C documentation into the docstrings?  It would make it more useful  
>>> in
>>> itself, and will be more accessible to programmers who don't know C
>>> that
>>> well.
>>
>> The README file explains where to find the C++ API docs (as opposed  
>> to
>> the C one, PyICU wraps ICU C++ classes) and how they correspond to  
>> the
>> python wrappers. It also explains how pythonic features, such as
>> iterators, string coversion, etc... are supported.
>> If you find holes or bugs in this file, please send them along.
>> As for copying the ICU docs, hmmm, patches are welcome.
>>
>>> Talking about PEPs - just checking:  do we want to keep the camel  
>>> case
>>> capitalisation to stay close to the C library?
>>
>> The wrapped class names and method names match exactly, where
>> possible, the ICU C++ ones. This is so that api docs can be matched
>> against, for example. The C api functions are _not_ wrapped by PyICU.
>>
>>> I might be able to help with some of this if I have a good example
>>> to go
>>> by. Is this something that everybody wants?
>>
>> What are you actually proposing ?
>>
>> Andi..
>
> Sorry about the C/C++ mixup.  What I am proposing is a way to ensure
> that "pydoc icu" will be useful and as self-contained as we can. I get
> docstrings in my editor for the libraries I use, so having the
> docstrings as complete as possible makes the python bindings even more
> useful.
>
> Someone who doesn't know C++ might find the C++ docs hard to read.   
> For
> example, the BreakIterator currently has the docstring  
> "t_breakiterator
> object", and the for its  __init__() I get
> x.__init__(...) initializes x; see x.__class__.__doc__ for signature
>
> t_breakiterator isn't even mentioned in the C++ docs, so this is  
> doubly
> confusing.

That's a bug in the macro I use to generate the Python C type. The  
docstring, which is worthless at the moment, should at least use the  
correct classname.

> I'm sure it is more work to do this, but will make the library more
> useful, so it is a suggestion, and I'm offering help to do some of  
> this.
> An example of how to do this (if it is agreed to be useful) will
> probably get me on my way quicker.

Ok, let me put an example together. This makes sense.

Andi..

>
> Regards
> Friedel
>
> --
> Recently on my blog:
> http://translate.org.za/blogs/friedel/en/content/provisionaly-done-localisation-guide-amharic
>
> _______________________________________________
> pyicu-dev mailing list
> pyicu-dev at osafoundation.org
> http://lists.osafoundation.org/cgi-bin/mailman/listinfo/pyicu-dev