The documentation within the tsearch2 backend labels it as being
experimental and that it should not be used. The have_backend() function
in roundup.backend returns False indicating that it does not exist and
is labeled as "currently not working". The PostgreSQL website also seems
to indicate that it has been deprecated since v8.3 when text searching
was integrated into the core[1].

Considering all this, it seems like the best option is to just remove
the tsearch2 backend.

[1] http://www.postgresql.org/docs/9.4/static/tsearch2.html

On 21/08/15 20:09, John Kristensen wrote:
>
> New submission from John Kristensen:
>
> The documentation within the tsearch2 backend labels it as being
> experimental and that it should not be used. The have_backend() function
> in roundup.backend returns False indicating that it does not exist and
> is labeled as "currently not working". The PostgreSQL website also seems
> to indicate that it has been deprecated since v8.3 when text searching
> was integrated into the core[1].
>
> Considering all this, it seems like the best option is to just remove
> the tsearch2 backend.
>
> [1] http://www.postgresql.org/docs/9.4/static/tsearch2.html

Unless someone can think of a good reason to keep the tsearch2 backend 
around I'll look at committing the patch to remove it sometime next week.

SeeYa,
John.

On Tue, Sep 22, 2015 at 5:19 PM, John Kristensen <john@jerrykan.com> wrote:

> On 21/08/15 20:09, John Kristensen wrote:
> >
> > New submission from John Kristensen:
> >
> > The documentation within the tsearch2 backend labels it as being
> > experimental and that it should not be used. The have_backend() function
> > in roundup.backend returns False indicating that it does not exist and
> > is labeled as "currently not working". The PostgreSQL website also seems
> > to indicate that it has been deprecated since v8.3 when text searching
> > was integrated into the core[1].
> >
> > Considering all this, it seems like the best option is to just remove
> > the tsearch2 backend.
> >
> > [1] http://www.postgresql.org/docs/9.4/static/tsearch2.html
>
> Unless someone can think of a good reason to keep the tsearch2 backend
> around I'll look at committing the patch to remove it sometime next week.
>

If I understand correctly, this is the backend that uses full text search
mechanism of PostgreSQL. Before removing that, it will be nice to have some
tests that prove that search works and some test for additional
capabilities of full text search that are being removed.

On 28/09/15 19:14, anatoly techtonik wrote:
> If I understand correctly, this is the backend that uses full text
> search mechanism of PostgreSQL. Before removing that, it will be nice to
> have some tests that prove that search works and some test for
> additional capabilities of full text search that are being removed.

I agree that more testing for the PostgreSQL backend would be good, but 
considering that the tsearch2 backend is essentially dead code that 
never seems to have made it to production quality, I don't think add 
tests should be a blocker to removing the backend.

SeeYa,
John.

On Fri, Oct 2, 2015 at 3:39 PM, John Kristensen <john@jerrykan.com> wrote:

> On 28/09/15 19:14, anatoly techtonik wrote:
>
>> If I understand correctly, this is the backend that uses full text
>> search mechanism of PostgreSQL. Before removing that, it will be nice to
>> have some tests that prove that search works and some test for
>> additional capabilities of full text search that are being removed.
>>
>
> I agree that more testing for the PostgreSQL backend would be good, but
> considering that the tsearch2 backend is essentially dead code that never
> seems to have made it to production quality, I don't think add tests should
> be a blocker to removing the backend.
>

I don't know if that code works or not - that's why I asked for tests.
Having search that finds at least some relevant info is better than no
search at all.

On 08/10/15 02:12, anatoly techtonik wrote:
> On Fri, Oct 2, 2015 at 3:39 PM, John Kristensen <john@jerrykan.com
> <mailto:john@jerrykan.com>> wrote:
>
>     On 28/09/15 19:14, anatoly techtonik wrote:
>
>         If I understand correctly, this is the backend that uses full text
>         search mechanism of PostgreSQL. Before removing that, it will be
>         nice to
>         have some tests that prove that search works and some test for
>         additional capabilities of full text search that are being removed.
>
>
>     I agree that more testing for the PostgreSQL backend would be good,
>     but considering that the tsearch2 backend is essentially dead code
>     that never seems to have made it to production quality, I don't
>     think add tests should be a blocker to removing the backend.
>
>
> I don't know if that code works or not - that's why I asked for tests.
> Having search that finds at least some relevant info is better than no
> search at all.

All of the tests currently in test/test_tsearch2.py have equivalent 
tests in test/test_postgresql.py

The tests for all the DB backends are essentially the same (the tests 
case classes all inherit tests from common test classes). If there is 
any new testing that is required, then those new tests are not just 
specific to the PostgreSQL backend, but would be needed for all DB backends.

Removing the tsearch2 backend is not going to leave the PostgreSQL 
backend with any less test coverage than the other DB backends. If there 
is a need for more tests, adding them should not be a blocker for 
removing the tsearch2 backend.

No one has raised a case for keeping the tsearch2 backend. Given that 
the tsearch2 extension has been abandoned and that it is unclear whether 
the backend was ever functional, I think it is safe to assume that we 
can get rid of it and remove some useless code from the project.

SeeYa,
John.

On Sat, Oct 10, 2015 at 5:58 AM, John Kristensen <john@jerrykan.com> wrote:

> Removing the tsearch2 backend is not going to leave the PostgreSQL backend
> with any less test coverage than the other DB backends.

I thought that tsearch2 is the only full text search capability for
PostgreSQL, no?

On 10/10/15 21:56, anatoly techtonik wrote:
> On Sat, Oct 10, 2015 at 5:58 AM, John Kristensen <john@jerrykan.com
> <mailto:john@jerrykan.com>> wrote:
>
>     Removing the tsearch2 backend is not going to leave the PostgreSQL
>     backend with any less test coverage than the other DB backends.
>
>
> I thought that tsearch2 is the only full text search capability for
> PostgreSQL, no?

 From my initial message in the issue:

"The PostgreSQL website also seems to indicate that [tsearch2] has been 
deprecated since v8.3 when text searching was integrated into the core[1]."

SeeYa,
John.

[1] http://www.postgresql.org/docs/9.4/static/tsearch2.html

On Sat, Oct 10, 2015 at 3:48 PM, John Kristensen <john@jerrykan.com> wrote:

> On 10/10/15 21:56, anatoly techtonik wrote:
>
>> On Sat, Oct 10, 2015 at 5:58 AM, John Kristensen <john@jerrykan.com
>> <mailto:john@jerrykan.com>> wrote:
>>
>>     Removing the tsearch2 backend is not going to leave the PostgreSQL
>>     backend with any less test coverage than the other DB backends.
>>
>>
>> I thought that tsearch2 is the only full text search capability for
>> PostgreSQL, no?
>>
>
> From my initial message in the issue:
>
> "The PostgreSQL website also seems to indicate that [tsearch2] has been
> deprecated since v8.3 when text searching was integrated into the core[1]."
>

Oh. Sorry. Go ahead then. =)

The tsearch2 backend has now been removed. I'll close this issue.

commit in reva86b0c02940d