Re: string

* David Binger wrote [2005-10-13 17:34:20 -0400]:

> 
> On Oct 13, 2005, at 5:00 PM, Michael Watkins wrote:
> >
> >string = either(str, unicode)
 
> This reminds me that I should preview upcoming changes related
> to spec and unicode.  I think we'll get a new release out next
> Tuesday evening.

I'm glad to hear that - cc'ing this to the Quixote list for Dulcinea fans
there who don't read the Durus list.

> We've put into the DulcineaPublisher constructor a statement
> that changes the quixote DEFAULT_CHARSET to 'utf8', so values
> we get from forms will be unicode.  You can't combine unicode
> with non-ascii str instances easily:  the automatic decoding
> can't work in that case.
> 
> So, we've added to a new SpecOperator named "string" to the spec
> module.  A "string" can be either a unicode or else an ascii str.
> The important feature of these strings is that they can be combined
> without risk of UnicodeEncodeErrors.  We changed every "str" attribute
> specification to "string" and ran an update script to convert all
> str attribute values to unicode, trying to guess the right charset for
> each instance (not always easy).
> 
> We also had to change some str() calls to quixote's stringify(),
> and there certain % operations need to be modified, but this
> conversion appears to be working as planned.

I'm not surprised to learn that your conversion wasn't instant but am glad to
see we'll all benefit from the effort.

I keep running into little unicode 'got-yahs' elsewhere but it does feel like
the whole ship is slowly moving there.