[Snowball-discuss] Out of date diffs?

J Smith jsmith@tutorbuddy.com
Thu Nov 28 00:42:01 2002


Greets, Martin,

Well, I figured out the problem with the stemmers. I was checking the ret=
urn=20
value of the stem() function and raising a notice if it returned 0. I ass=
umed=20
a return value of 0 meant some kind of error, I guess. The problem was, I=
=20
didn't notice the error being raised as the PHP script used for testing=20
ignored errors of type E_NOTICE, which was being thrown. After setting th=
e=20
error reporting level to catch all errors, the problem was obvious, as to=
ns=20
of notices started popping up during the test script's execution.

So, long story short, *_stem() returning 0 does not indicate an error.

Sorry for raising a flag unnecessarily. This was all me.

J





On November 27, 2002 04:30 am, Martin Porter wrote:
> J,
>
> Hello there!
>
> I'm not aware of any inconsistencies of the type you report, and am
> doubtful if they are real, since nothing has changed on the site in qui=
te a
> while, and people are usual quick to pick up these things.
>
> Let's try and pin it down: as I see things 'a' heads the vocab.txt file=
 and
> output.file of 'english' (porter2), and the english stemmer stems 'a' t=
o
> 'a'. None of the stemmers actually stem a non-null string to a null str=
ing.
> What exactly are you getting here?
>
> Martin