[Xapian-tickets] [Xapian] #346: Python 3 support

[Xapian]

#346: Python 3 support
--------------------------------------+------------------------------
 Reporter:  olly                      |             Owner:  richard
     Type:  defect                    |            Status:  assigned
 Priority:  highest                   |         Milestone:  1.3.2
Component:  Xapian-bindings (Python)  |           Version:  SVN trunk
 Severity:  normal                    |        Resolution:
 Keywords:                            |        Blocked By:
 Blocking:                            |  Operating System:  All
--------------------------------------+------------------------------
\
\
\
\
\
\

Comment (by james):

 I'm +1 on defaulting to using bytes, and +1 talking about "string" methods
 (whether `_str`, `_s`; I'm indifferent between them) instead of "unicode",
 since this is Py3.

 I wonder if `TermGenerator` and `QueryParser` would ideally be namespace-
 separated (`xapian.termparsing` or something), since they go together and
 are an optional higher-level interface.

 I wonder separately if since it doesn't affect stuff stored on disk,
 having `query_parser.set_prefix` accept both `bytes` and `str` types and
 serialising `str` -> `bytes` using an extra parameter (which defaults to
 `'utf8'`) might not be a good solution to that problem. I'm guessing that
 most people are going to want to think in terms of strings across the
 board; if you're just dealing in unicode data and are happy with UTF-8
 (which is probably most Western applications?) then not having to think
 about bytes serialisation at all is a lot clearer.

 I also wonder still if having an explicit wrapper (`from xapian.utf8
 import Document, Database`) bundled as part of our own bindings which uses
 `str` everywhere that you'd usually have `bytes` and a `_s` variant, might
 stop people writing almost-trivial wrappers. Although honestly if all `_s`
 methods and functions take only `str` and return only `str`, with
 conversion in and out via `utf8`, then we're probably fine for new
 applications, having a `xapian.utf8` wrapper set built-in would be useful
 in porting from Py2. (An alternative would be something like `2to3` for
 Xapian client code.)
\
\
\

--
Ticket URL: <http://trac.xapian.org/ticket/346#comment:56>
Xapian <http://xapian.org/>
Xapian