Re: [PATCH] Improve geometric types

Hello, sorry to be late.

At Fri, 2 Mar 2018 15:26:25 +0100, Emre Hasegeli <emre(at)hasegeli(dot)com> wrote in <CAE2gYzzj6wgGeBY1vdAHBzzQDSDs-8NhD+=eTcgBKEsCBWpUXg(at)mail(dot)gmail(dot)com>
> > Unfortunately according to http://commitfest.cputube.org/ this patch doesn't apply anymore.
> New versions are attached.

Thank you for the revised version. This seems fine for me so
marked this as "Ready for Committer".

- This applies cleanly on the master head and regression passes.

- The new behavior looks sane (except for the EPSILON, which is
out of the scope of this patch).

- Test is complete as long as I can see. At least far more
completely filled than the previous state. Some test items
might seem a bit big, but it seems to be needed to raise
coverage on required combinaions of dimension values.

By the way I think that line_perp is back patched, could you
propose it for the older versions? (I already marked my entry as
Rejected)

Brief summary follows (almost same to the header of patch files):

- 0001 looks fine.
> * Eliminate SQL level function calls
> * Re-use more functions to implement others
> * Unify internal function names and signatures
> * Remove private functions from geo_decls.h
> * Replace not-possible checks with Assert()
> * Add comments to describe some functions
> * Remove some unreachable code
> * Define delimiter symbols of line datatype like the other ones
> * Remove debugging print()s from the refactored functions

And one bug fix.

- 0002 looks fine.

Refactors adt/float.c and utils/float.h
making float checking *macros* into inline functions.
making float comparison operators more efficiently.

others are the consequence of the above change.
and fixes NaN problem of GiST.

- 0003 looks fine.

just changes the usage of double to float8 as more proper type.
uses operator functions instead of bare arithmetics to handle
arithmetic errors more properly.

- 0004 looks fine. (Sorry for overlooking that this treats bugs)

all overlooked failure cases seems to be fixed.

- 0005 looks fine.

this unifies +-0.0 to +0.0 for the convenient of later processing.

- 0006 It seems cover the all types of operations.

regards,

--
Kyotaro Horiguchi
NTT Open Source Software Center