I just took a first crack at them:

    http://wiki.apache.org/solr/ReleaseNote34
    http://wiki.apache.org/lucene-java/ReleaseNote34

Feel free to go improve them before we release!

Mike McCandless

http://blog.mikemccandless.com

---------------------------------------------------------------------

Will the fix/patch for issue SOLR-2726 included in SOLR 3.4.0?
----- Ursprüngliche Nachricht -----
Von: Michael McCandless <[EMAIL PROTECTED]>
Datum: Freitag, 9. September 2011, 20:50
Betreff: 3.4.0 draft release notes
An: Lucene/Solr dev <[EMAIL PROTECTED]>

> I just took a first crack at them:
>
>     http://wiki.apache.org/solr/ReleaseNote34
>     http://wiki.apache.org/lucene-java/ReleaseNote34
>
> Feel free to go improve them before we release!
>
> Mike McCandless
>
> http://blog.mikemccandless.com
>
> -----------------------------------------------------------------
> ----
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>

---------------------------------------------------------------------

On Sat, Sep 10, 2011 at 10:21 AM, Bernd Fehling
<[EMAIL PROTECTED]> wrote:

> Will the fix/patch for issue SOLR-2726 included in SOLR 3.4.0?

Sorry, no.

This isn't a release blocker issue.

But, separately, I think we should fix it, but on quick glance it
doesn't look like there's consensus on how to fix it?

Mike McCandless

http://blog.mikemccandless.com

---------------------------------------------------------------------

On Sun, Sep 11, 2011 at 7:04 PM, Michael McCandless
<[EMAIL PROTECTED]> wrote:
> On Sat, Sep 10, 2011 at 10:21 AM, Bernd Fehling
> <[EMAIL PROTECTED]> wrote:
>
>> Will the fix/patch for issue SOLR-2726 included in SOLR 3.4.0?
>
> Sorry, no.
>
> This isn't a release blocker issue.
>
> But, separately, I think we should fix it, but on quick glance it
> doesn't look like there's consensus on how to fix it?
>

I had this same bug when implementing a spellchecker too.
Its something the spellcheck framework expects, but doesn't provide.

I think its broken that SolrSpellChecker has both field name and analyzer,
but only sets up field name in its init()... if SolrSpellChecker is
going to own the 'analyzer' variable then
I think its init() should take care of the logic, currently its either
duplicated across spellchecker implementations,
or its missing entirely, causing bugs like SOLR-2726.
--
lucidimagination.com

---------------------------------------------------------------------

Hmm, it may be not an blocker issue but still has a NPE which isn't
good publicity for solr/lucene.
This is already fixed with a patch attached to SOLR-2726.
And yes, it is fixed with its init().
So where is the problem?

----- Ursprüngliche Nachricht -----
Von: Robert Muir <[EMAIL PROTECTED]>
Datum: Montag, 12. September 2011, 1:13
Betreff: Re: 3.4.0 draft release notes
An: [EMAIL PROTECTED]

> On Sun, Sep 11, 2011 at 7:04 PM, Michael McCandless
> <[EMAIL PROTECTED]> wrote:
> > On Sat, Sep 10, 2011 at 10:21 AM, Bernd Fehling
> > <[EMAIL PROTECTED]> wrote:
> >
> >> Will the fix/patch for issue SOLR-2726 included in SOLR 3.4.0?
> >
> > Sorry, no.
> >
> > This isn't a release blocker issue.
> >
> > But, separately, I think we should fix it, but on quick glance it
> > doesn't look like there's consensus on how to fix it?
> >
>
> I had this same bug when implementing a spellchecker too.
> Its something the spellcheck framework expects, but doesn't provide.
>
> I think its broken that SolrSpellChecker has both field name and
> analyzer,but only sets up field name in its init()... if
> SolrSpellChecker is
> going to own the 'analyzer' variable then
> I think its init() should take care of the logic, currently its either
> duplicated across spellchecker implementations,
> or its missing entirely, causing bugs like SOLR-2726.
>
>
> --
> lucidimagination.com
>
> -----------------------------------------------------------------
> ----
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>

---------------------------------------------------------------------

On Tue, Sep 13, 2011 at 2:53 PM, Bernd Fehling
<[EMAIL PROTECTED]> wrote:
> Hmm, it may be not an blocker issue but still has a NPE which isn't
> good publicity for solr/lucene.
> This is already fixed with a patch attached to SOLR-2726.
> And yes, it is fixed with its init().
> So where is the problem?

As i explained both in this thread and on the jira issue, the proper
fix is for the logic to be in *SolrSpellChecker*, since its the one
that has this 'analyzer' variable.

then the logic doesnt need to be duplicated across n implementing
classes (AbstractLuceneSpellChecker, DirectSpellChecker, Suggester,
...)

--
lucidimagination.com

---------------------------------------------------------------------