[postgis-devel] EOL 2.2 after 2.2.7

Fri Mar 30 11:45:20 PDT 2018

You are right of course Regina. I was mixing postgres versions when trying to understand. Sorry

And I fully agree that maintaining postgis 2.2 for very long isn't possible.

Waiting for 2.5 might be good as you say though. 

Thanks
Nicklas

Sent from my Sony Xperia™ smartphone

---- Regina Obe wrote ----

>Sorry to tell you this Nicklas, but PostGIS 2.2 came out around the same time as PostgreSQL 9.5.
>
>Remember we were sporting this new feature called true-KNN that required PostgreSQL 9.5 to take advantage of it.
>
> 
>
>When 9.3 came out we were still on 2.1, 2.2 took a long time to come out.
>
>We were still on 2.1 when 9.4 came out in (2014-12-18)
>
> 
>
>So with that theory we couldn't EOL'd 2.2 till 9.5 is EOL'd, EOL January 2021 way too long.
>
> 
>
>True some packagers started offering the 2.3 even for lower PostgreSQL like 9.4 when 2.3 out, but that was usually from backports.
>
>I think the main ports that had 9.4 are stuck on 2.1.
>
>And I really want to encourage packagers to carry newer versions.  What better way than to deprecate 2.2.
>
> 
>
>My thinking for saying at the very least the first PostgreSQL (I guess to be clear I should say lowest PostgreSQL version)
>
>supported by a PostGIS minor, is so we don't have to keep these 
>
>ancient versions around and testing them just because we missed a PostgreSQL cycle or 2.
>
> 
>
>The main mindset that security patches we could possibly put in these older, are not the critical bug fixes people would be looking for
>
>so it's of minimal value to users to keep these going and major hindrance to us to keep on patching and testing them.  If it is not the lowest PostgreSQL for a PostGIS supported, then that means there is at least one newer PostGIS minor they can upgrade to.
>
> 
>
>If they really wanted those things, they have a safe migration path to a newer PostGIS minor without having to do a dump / restore of their database.  Granted they might have to compile the PostGIS themselves, but that's their problem for waiting so long to upgrade PostgreSQL.  
>
> 
>
>For example things like making PostGIS schema non-relocatable so that we could schema qualify all functions is not something
>
>we can backport to 2.2 as it's a major policy change.  But IS something a lot of people running 2.1 and 2.2 want so their restores restore cleanly and their materialized views can be refreshed.
>
> 
>
>All that said.  I'm thinking perhaps it is a little too soon to EOL 2.2.  So we can wait another 3 months or when PostGIS 2.5 comes out.
>
> 
>
>Thanks,
>
>Regina
>
> 
>
>From: postgis-devel [mailto:postgis-devel-bounces at lists.osgeo.org] On Behalf Of Nicklas Aven
>Sent: Friday, March 30, 2018 10:23 AM
>To: Darafei Kom?pa Praliaskouski <me at komzpa.net>
>Cc: PostGIS Development Discussion <postgis-devel at lists.osgeo.org>
>Subject: Re: [postgis-devel] EOL 2.2 after 2.2.7
>
> 
>
>I think you misunderstood mi slightly. 
>When oostgis 2.2 was released the latest postgresql version was 9.3. If people installed postgis on something older than 9.3 by then cannot be a problem of postgis project. Also if they installed some old postgis version back then it is up to them, but not the problem of Postgis. 
>
>So, one way to argue is to support until EOL of 9.3. But since it also a matter of work to do and pther practicalbthings I have no problem with ending 2.2 earlier. But if following postgres versions I think 9.3 would make most sense.
>
>/Nicklas
>
>Sent from my Sony Xperia™ smartphone
>
>---- Darafei "Komяpa" Praliaskouski wrote ----
>
>In that logic:
>
> 
>
>Postgres 9.2 (EOL September 2017) had PostGIS 2.1 to 2.3 as latest at various points in time,
>
>Postgres 9.3 (EOL September 2018, still alive) had PostGIS 2.1 to 2.4 as latest,
>
>Postgres 9.4 (EOL December 2019) had PostGIS 2.1 and supported until trunk,
>
>Postgres 9.5 (EOL January 2021) had PostGIS 2.2 and supported until trunk.
>
> 
>
>Does it mean we can't EOL 2.1 until December 2019, and can't EOL 2.2 until Jan 2021?
>
> 
>
>I think a simple rule "we support versions enough to cover all non-EOL Postgres versions, at least two of them" is more like what we're trying to say with all the "2-4 years" and "until {newest, oldest} postgres version is EOL". 
>
> 
>
>пт, 30 мар. 2018 г. в 16:08, Nicklas Aven <nicklas.aven at jordogskog.no <mailto:nicklas.aven at jordogskog.no> >:
>
>As someone said before. When you do the job to upgrade postgresql major version you probably also upgrade postgis. Then, from a user perspective, you probably expect Postgis to live as long as the newest postgresql version at that point. That means that when people upgraded to pg 9.3 they installed postgis  2.2 and probably expect postgis to be supported until they are forced to upgrade postgresql.
>
>What the minimum version of postgresql, postgis supports I think anyway  is irrelevant.
>
>Regards
>Nicklas
>
>Sent from my Sony Xperia™ smartphone
>
>---- Darafei "Komяpa" Praliaskouski wrote ----
>
>Given that we depend on minimal Postgres versions in our decisions, can we give EOL dates for each PostGIS release in advance then?
>
> 
>
>Basically make a join of 
>
>https://www.postgresql.org/support/versioning/ 
>
>and
>
>https://trac.osgeo.org/postgis/wiki/UsersWikiPostgreSQLPostGIS
>
> 
>
>This way EOL for 2.2 is EOL of 9.1  -  September 2016
>EOL for 2.3 is EOL of 9.2 - September 2017
>EOL for 2.4 is EOL of 9.3 - September 2018
>
> 
>
>This way 2.4 lives just for a year though and EOLs with a release of 2.5 - is that a desired thing? 
>
>If 2.5 accidentially supports 9.3, shall we EOL it right at release?
>
> 
>
>пт, 30 мар. 2018 г. в 0:55, Regina Obe <lr at pcorp.us <mailto:lr at pcorp.us> >:
>
>Yah Darafei mentioned that in IRC.  I guess I've never been good at elementary math.
>
> 
>
>Well I suppose I could bare another release of 2.2, or we change it to 2-4 years or whenever the earliest supported version of a PostgreSQL on a release reaches EOL.
>
> 
>
>So since 2.2 first release supported is 9.1 we can EOL it now. Since 9.1 is already EOL'd
>
> 
>
> 
>
> 
>
>From: postgis-devel [mailto:postgis-devel-bounces at lists.osgeo.org <mailto:postgis-devel-bounces at lists.osgeo.org> ] On Behalf Of Daniel Baston
>Sent: Thursday, March 29, 2018 4:51 PM
>To: PostGIS Development Discussion <postgis-devel at lists.osgeo.org <mailto:postgis-devel at lists.osgeo.org> >
>Subject: Re: [postgis-devel] EOL 2.2 after 2.2.7 <tel:2.2.7> 
>
> 
>
>The policy on postgis.net <http://postgis.net>  suggests we should support 2.2 for 3-4 years,
>
>i.e. until sometime between October 2018 -  October 2019.
>
> 
>
>Dan
>
> 
>
>On Tue, Mar 27, 2018 at 1:42 PM, Regina Obe <lr at pcorp.us <mailto:lr at pcorp.us> > wrote:
>
>I thought we had EOL'd 2.2, but guess not since we are still testing it and
>never made an official EOL announcement
>
>Everyone okay if I do one more release of 2.2, and then EOL it and take it
>off our testing menu?
>
>Thanks,
>Regina
>
>_______________________________________________
>postgis-devel mailing list
>postgis-devel at lists.osgeo.org <mailto:postgis-devel at lists.osgeo.org> 
>https://lists.osgeo.org/mailman/listinfo/postgis-devel
>
> 
>
>_______________________________________________
>postgis-devel mailing list
>postgis-devel at lists.osgeo.org <mailto:postgis-devel at lists.osgeo.org> 
>https://lists.osgeo.org/mailman/listinfo/postgis-devel
>
>
>_______________________________________________
>postgis-devel mailing list
>postgis-devel at lists.osgeo.org
>https://lists.osgeo.org/mailman/listinfo/postgis-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20180330/9f432378/attachment-0001.html>