Re: [documancer] ANN: new Documancer release

[Arnd B. <arn...@we...>]

Hi Vaclav,

thanks for your rapid response!


On Tue, 15 Feb 2005, Vaclav Slavik wrote:

> Hi,
>
> Arnd Baecker wrote:
> > This is with python 2.3.4, wxmozilla-0.5.3, documancer 0.2.5.
>
> Just a guess: what is your locale and what is decimal point deliminer
> used in it? Does running it with LC_ALL=en_GB or =C help?

env shows that LANG=de_DE.UTF-8
With  export  LC_ALL=en_GB everything seems to work fine!

((Note that without this setting for example

I played around with the new search a bit more:
- searching for "lists dafdsf" (in the python 2.3.4 docs)
  also gives not error and the expected hits
  (the second word is just garbage)
))

Do you think that it is possible that documancer
reacts a bit more "helpful" when one has the wrong
locale settings?
(this is not too important for me but helpful as the fix is easy
to put in a script which calls documancer,
but some users might start documancer directly and
get into trouble)

> > BTW: creating the index seems (if my memory serves me correctly)
> > to be much slower than with swish.
>
> Yes, that's true. But I think it's worth it -- no other fulltext
> search I evaluated, SWISH-E included, supports Unicode. And Lucene's
> API is really useful...

OK, so this makes it even more important
to supply pre-generated indizes system wide
(as we did last year for our students).

Many thanks,

Arnd