Re: Inquiry on planned PostGIS releases and minor-version behavior (now–Feb 28)

[Paul Ramsey]

On Wed, Jan 21, 2026 at 10:36 AM Bharath Tadikonda <
tadikondabharath at gmail.com> wrote:

> Are there any planned PostGIS releases (minor or patch) between now and
February 28?

No, but that doesn't mean they won't happen. Minor releases expect in the
fall / late summer. Patch releases expect any time we feel the urge, either
when enough fixes have piled up, or a particularly nasty bug has been
repaired.

> If not, when is the next postgis version release anticipated?

Cannot say, except for what is stated above.

> For PostGIS minor/patch releases (e.g., 3.6.x → 3.6.y):
> Are changes strictly bug fixes, or can they include behavioral changes to
existing functions?

No deliberate behaviour change, except if existing behaviour is wrong.
At any time, if your provider updates the GEOS or Proj library at more than
a patch level, that could cause behaviour changes in PostGIS, so you must
watch both PostGIS and the major libraries it uses.

> Are there specific classes of functions (geometry processing, topology,
raster, etc.) where behavior changes are more likely?

Not really.

> Release notes & compatibility guidance:
> Is there a recommended way to identify function-level behavior changes
between minor versions (beyond high-level release notes)?

Nope.

> Are there guarantees or best-practice guidance around backward
compatibility expectations for PostGIS patch releases?

As above, we strive to follow semver guarantees. Bug fixes only on patch.
Backward compatible changes and new features only on minor. Backwards
incompatibilities restricted to major.

> Operational best practices:
> Do you recommend pinning PostGIS patch versions in production where
possible, or validating only specific functions we rely on most during
upgrades?

No, I recommend always having the latest patch release at all times for
PostGIS, GEOS, Proj. I recommend regularly testing and pushing GEOS and
PostGIS forward in minor versions as often as possible, as performance and
correctness improvements are always landing. Many of the tickets that are
reported to us are just people with old pinned versions, whose problems
were long since fixed. Only relatively small bugs get back-ported, large
improvements in behaviour and performance (like RelateNG or OverlayNG) are
too big to back-port, and only available through upgrade.

> Any guidance you can share will help us reduce risk and plan upgrades
more confidently.

I'm no ops genius. Hope the above is useful.

ATB,

P

>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20260121/68a69aa5/attachment.htm>