Horton

Re: [PATCH] Improve geometric types

Robert Haas <robertmhaas@gmail.com>

View thread View on postgr.es
From: Robert Haas <robertmhaas@gmail.com>
To: Kyotaro HORIGUCHI <horiguchi.kyotaro@lab.ntt.co.jp>
Cc: Emre Hasegeli <emre@hasegeli.com>, Aleksander Alekseev <a.alekseev@postgrespro.ru>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2017-10-02T12:23:49Z
Lists: pgsql-hackers 
On Mon, Oct 2, 2017 at 4:23 AM, Kyotaro HORIGUCHI
<horiguchi.kyotaro@lab.ntt.co.jp> wrote:
> For other potential reviewers:
>
> I found the origin of the function here.
>
> https://www.postgresql.org/message-id/4A90BD76.7070804@netspace.net.au
> https://www.postgresql.org/message-id/AANLkTim4cHELcGPf5w7Zd43_dQi_2RJ_b5_F_idSSbZI%40mail.gmail.com
>
> And the reason for pg_hypot is seen here.
>
> https://www.postgresql.org/message-id/407d949e0908222139t35ad3ad2q3e6b15646a27dd64@mail.gmail.com
>
> I think the replacement is now acceptable according to the discussion.
> ======

I think if we're going to do this it should be separated out as its
own patch.  Also, I think someone should explain what the reasoning
behind the change is.  Do we, for example, foresee that the built-in
code might be faster or less likely to overflow?  Because we're
clearly taking a risk -- most trivially, that the BF will break, or
more seriously, that some machines will have versions of this function
that don't actually behave quite the same.

That brings up a related point.  How good is our test case coverage
for hypot(), especially in strange corner cases, like this one
mentioned in pg_hypot()'s comment:

 * This implementation conforms to IEEE Std 1003.1 and GLIBC, in that the
 * case of hypot(inf,nan) results in INF, and not NAN.

I'm potentially willing to commit a patch that just makes the
pg_hypot() -> hypot() change and does nothing else, if there are not
objections to that change, but I want to be sure that we'll know right
away if that turns out to break.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Commits

  1. Improve test coverage of geometric types

  2. Fix problems in handling the line data type

  3. Use the built-in float datatypes to implement geometric types

  4. Remove remaining GEODEBUG references from geo_ops.c

  5. Provide separate header file for built-in float types

  6. Refactor geometric functions and operators

  7. Fix crash in close_ps() for NaN input coordinates.

  8. Fix GiST index build for NaN values in geometric types.

  9. Enable building with Visual Studion 2013.

  10. Suppress -0 in the C field of lines computed by line_construct_pts().