[postgis-tickets] r16188 - Weight-aware ST_GeometricMedian

[Darafei "Komяpa" Praliaskouski]

Hi Dan,

I'm sorry to have committed that change too fast.
I was under impression that I've addresses all of the comments you've made
on the code. Now I see that I replied not to all of them, but just had
those hidden by github as underlying code changed according to the comment,
and shown only in PR history.

I've reopened PR so we can continue with review. I hope we can do that
without a revert. (I need the math there to work well too, so it's not in
my interest to ignore issues :)

I'll start thread on size_t for counters on -dev.

ср, 27 дек. 2017 г. в 17:25, Daniel Baston <dbaston at gmail.com>:

> Given that this approaches a complete rewrite, I would have appreciated
> the chance to review this before you committed it to trunk. The PR
> currently has several open questions. This commit also makes some style
> changes (size_t instead of int for counters is one) that should be brought
> up on the dev list, so we can get a ticket to apply them globally if they
> are desirable.
>
> Dan
>
> On Tue, Dec 26, 2017 at 9:07 AM, Darafei <komzpa at gmail.com> wrote:
>
>> Author: komzpa
>> Date: 2017-12-26 06:07:49 -0800 (Tue, 26 Dec 2017)
>> New Revision: 16188
>>
>> Modified:
>>    trunk/NEWS
>>    trunk/doc/reference_measure.xml
>>    trunk/liblwgeom/cunit/cu_algorithm.c
>>    trunk/liblwgeom/lwgeom_median.c
>> Log:
>> Weight-aware ST_GeometricMedian
>>
>> Weight is to be supplied as M ordinate of individual points in MultiPoint.
>>
>> Bring lwgeom_median.c cunit test coverage to 100%.
>>
>> ST_GeometricMedian(fail_if_not_converged=false, max_iter=0)
>> might give you weighted centroid, but that's not a promise.
>>
>> Closes #3954
>> Closes https://github.com/postgis/postgis/pull/176
>>
>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-tickets/attachments/20171227/4ecf5b7b/attachment-0001.html>