[postgis-users] "value out of range: underflow" blocking error

Marco Boeringa marco at boeringa.demon.nl
Thu Feb 10 12:51:35 PST 2022


Hi Regina,

Thanks very much for this useful information. Yes, I appreciate there 
may be breaking changes causing trouble.

Maybe it would be good though, to provide some basic guidance on the 
PostGIS webpages regarding all of the different dependencies that 
"SELECT PostGIS_Full_Version()" by default lists (GEOS, PROJ, WAYGU), 
and whether or not it is possible and safe to "upgrade" such 
dependencies without breaking things.

E.g. the reason I attempted this at all and add the 'ubuntugis' PPA, is 
related to all the nice performance and functionality enhancements I saw 
listed for "GEOS 3.10.1", and that Paul blogged about quite extensively, 
versus the older "3.8.0" that I had installed. Of course, you made clear 
that any of the newer functionality, e.g. new functions, is not 
available, but the performance enhancements should still apply, because 
the underlying machinery changed. Am I right here?

So in any case, it probably is only useful and sensible ever to attempt 
a GEOS update based on your remarks, and likely no other updates are 
possible unless upgrading the whole set via a custom re-compilation of 
entire PostGIS.

Marco

Op 10-2-2022 om 21:15 schreef Regina Obe:
>
> Yes. That version is telling you that PostGIS was compiled with PROJ 
> 6.3.1.
>
> The check is done at compile time cause we have a lot of IF DEFS to 
> handle various proj changes.
>
> It’s less of an issue with GEOS since with GEOS just means you will 
> not be getting the newer features.
>
> For PROJ it often means misbehavior since I don’t think the major 
> versions are completely backward compatible with older.
>
> *From:*postgis-users [mailto:postgis-users-bounces at lists.osgeo.org] 
> *On Behalf Of *Marco Boeringa
> *Sent:* Thursday, February 10, 2022 2:08 AM
> *To:* postgis-users at lists.osgeo.org
> *Subject:* Re: [postgis-users] "value out of range: underflow" 
> blocking error
>
> Hi Regina,
>
> This was over-optimistic.
>
> I now again see the "value out of range: underflow" error. However, 
> this now happened after upgrade of PROJ 6.3.1 to 8.2.0. Note that last 
> time, I also did this upgrade, as it was a forced and automatic one 
> suggested by the Ubuntu "Software Updater". So in both cases after 
> restoring my VM, this update from PROJ 6.3.1 to 8.2.0 happened before 
> the issue started appearing. So it appears the issue is related to the 
> PROJ update, not the GEOS update as I initially suspected.
>
> One interesting thing I noted, is that the postgis_full_version still 
> shows 'PROJ="6.3.1"', instead of 8.2.0, despite the Synaptic Package 
> Manager clearly showing only 8.2.0 is installed:
>
> PostGIS version: POSTGIS="3.2.0 c3e3cc0" [EXTENSION] PGSQL="130" 
> GEOS="3.10.1-CAPI-1.16.0" PROJ="6.3.1" LIBXML="2.9.10" 
> LIBJSON="0.13.1" LIBPROTOBUF="1.3.3" WAGYU="0.5.0 (Internal)"
>
> Now I remember a similar issue with outdated version number for GEOS 
> from last year when I attempted to update GEOS from 3.6.0 to 3.8.0, 
> where Paul remarked something like that it meant that my PostGIS 
> "wasn't compiled with GEOS 3.8.0 support", so does this possibly also 
> mean my current PostGIS 3.2.0 installed from the official Ubuntu 
> repository is also not compiled with PROJ 8.2.0 support, and I should 
> downgrade to 6.3.1?
>
> Marco
>
> Op 9-2-2022 om 12:57 schreef Marco Boeringa:
>
>     Hi Regina,
>
>     After restoring my VM a second time, I dug a little deeper into
>     the problem, and I now think I understand what went wrong:
>
>     I had initially attempted to upgrade my GEOS 3.8.0 that comes by
>     default in the Ubuntu 20.04 install to GEOS 3.10.1 from the
>     'ubuntugis-unstable' PPA.
>
>     To do this, after adding the PPA in Ubuntu, I checked the
>     'libgeos3.10.1' entry in the Synaptic Package Manager, and clicked
>     "Apply". This succeeded, but I did not uninstall 'libgeos3.8.0' in
>     the same process, so I had two 'libgeos' versions installed.
>
>     This seems to have caused the issue, as after I now uninstalled
>     'libgeos3.8.0', I noticed the Synaptic Package manager updated a
>     couple of dependent libraries to the 3.10.1 version:
>     'libgeos++dev','libgeos-c1v5','libgeos-dev'. So I guess the issue
>     with the underflow was caused by having the outdated 3.8.0
>     versions of these libraries still installed.
>
>     Marco
>
>     Op 8-2-2022 om 19:23 schreef Regina Obe:
>
>         You have the create table statement we can test with?
>
>
>             -----Original Message-----
>             From: postgis-users
>             [mailto:postgis-users-bounces at lists.osgeo.org
>             <mailto:postgis-users-bounces at lists.osgeo.org>] On
>             Behalf Of Marco Boeringa
>             Sent: Monday, February 7, 2022 2:34 PM
>             To: postgis-users at lists.osgeo.org
>             Subject: Re: [postgis-users] "value out of range:
>             underflow" blocking
>
>         error
>
>             One additional remark: in my case, there is no involvement
>             of a GiST index
>
>         as
>
>             per the PostgresPro link. It fails in a "CREATE TABLE"
>             type statement that
>
>         also
>
>             includes some calculations, so there isn't any indexes yet
>             at that stage.
>
>             Op 7-2-2022 om 08:17 schreef Marco Boeringa:
>
>                 Hi,
>
>                 Anyone else seeing this issue pop up, possibly after a
>                 recent update
>                 to a PostgreSQL related component, or something that
>                 PostgreSQL /
>                 PostGIS is dependent on?
>
>                 I am still not sure if it is caused by something in my
>                 own coding, and
>                 I first though this might be related to an attempt to
>                 upgrade to
>                 PostgreSQL 14, but after restoring my machine to
>                 PostgreSQL 13(.5), I
>                 still see this error blocking the successful execution
>                 of a crucial
>                 query in my toolchain, that used to run fine up until
>                 recently.
>
>                 I noticed there has been some very recent discussion
>                 about this same
>                 type of error on a PostgreSQL mailing list:
>
>                 https://postgrespro.ru/list/thread-id/2580248
>
>                 Marco
>
>                 _______________________________________________
>                 postgis-users mailing list
>                 postgis-users at lists.osgeo.org
>                 https://lists.osgeo.org/mailman/listinfo/postgis-users
>
>             _______________________________________________
>             postgis-users mailing list
>             postgis-users at lists.osgeo.org
>             https://lists.osgeo.org/mailman/listinfo/postgis-users
>
>         _______________________________________________
>         postgis-users mailing list
>         postgis-users at lists.osgeo.org
>         https://lists.osgeo.org/mailman/listinfo/postgis-users
>
>     _______________________________________________
>     postgis-users mailing list
>     postgis-users at lists.osgeo.org
>     https://lists.osgeo.org/mailman/listinfo/postgis-users
>
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/postgis-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20220210/8109a750/attachment.html>


More information about the postgis-users mailing list