Re: NotImplementedError: Postgres FTS not available

Hi Martin:

In message <[email protected]>,
=?UTF-8?Q?Martin_=c3=96stlund?= via Roundup-users writes:
>Following along, the Upgrade documentation for 2.2.0 to enable 
>native-fts for Postgres at
>https://www.roundup-tracker.org/docs/upgrading.html#enhanced-full-text-search-optional
>
>Changing indexer to native-fts in config.ini (and running a supported 
>PostgreSQL), reindexing my database throws an exception:

I hope your upgrade to 2.2.0 wasn't to just use Postgres FTS. There
are a few fixes needed for Postgres FTS to work. They are in the
current development trunk. It turns out my test cases for 2.2.0's
PostgreSQL FTS were not testing what I thought they were. In fixing
that I found a few more issues.

  - Enable postgres-fts: fix indexer-common::get_indexer so it returns a
    postgresql-fts. Test code paths in get_indexer. (John Rouillard)
  - Fix Postgres native-fts, implement a two phase initialization of the
    indexer. The native-fts one gets assigned after the database
    connection is open. (John Rouillard)
  - fix crash if postgresql native-fts backend is asked to index content
    with null bytes. (John Rouillard)

The one on top fixes:

>line 150, in get_indexer
>     raise NotImplementedError("Postgres FTS not available")
>NotImplementedError: Postgres FTS not available
>[roundup@intlab roundup-2.2.0]$
>
>149         if db.dbtype == "postgres":
>150             raise NotImplementedError("Postgres FTS not available")
>151             from roundup.backends.indexer_postgres_fts import Indexer
>152             return Indexer(db)

>Should it really raise an NotImplementedError here?

Umm, no it shouldn't have. It should have raised a BuggyImplementation
exception 8-). That is more appropriate given the other bugs I found
8-/.  Sorry.
 
If you really need Postgresql FTS, you can run the development
branch. There is no way to cherrypick the commits reliably. It would
be good to get somebody to test it in a real use case. Sadly nobody
tried it during the beta period. My "production" trackers run on
sqlite with whoosh so...

After fixing the tests and things, I did spin up a tracker and tested
it manually against a local (14.x) postgres. It seemed to work. I was
able to get the proper issues returned (using web and tsquery type
searches) and the native search __words table was empty. It didn't get
a lot of testing with edge cases etc. trying to break it.

I also just realized that the Errata page
(https://wiki.roundup-tracker.org/ReleaseErrata) isn't updated for
this. I'll fix that.

Sorry I don't have better news. Have a great weekend.

--
				-- rouilj
John Rouillard
===========================================================================
My employers don't acknowledge my existence much less my opinions.