Received: from malur.postgresql.org ([217.196.149.56]) by arkaria.postgresql.org with esmtps (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.98.2) (envelope-from ) id 1xOYvy-00000000D7y-1GC6 for pgsql-pkg-yum@arkaria.postgresql.org; Wed, 04 Nov 2026 11:13:39 +0000 Received: from localhost ([127.0.0.1]:54842 helo=malur.postgresql.org) by malur.postgresql.org with esmtp (Exim 4.98.2) (envelope-from ) id 1xOYvy-00000005xIv-33U5 for pgsql-pkg-yum@arkaria.postgresql.org; Wed, 04 Nov 2026 11:13:38 +0000 Received: from magus.postgresql.org ([2a02:c0:301:0:ffff::29]:45062) by malur.postgresql.org with esmtps (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.98.2) (envelope-from ) id 1xOYvy-00000005xIm-2530 for pgsql-pkg-yum@lists.postgresql.org; Wed, 04 Nov 2026 11:13:38 +0000 Received: from ma04-relay.lansolnet.com ([176.95.46.38]) by magus.postgresql.org with esmtps (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.98.2) (envelope-from ) id 1xOYvw-00000000CPC-0mK4 for pgsql-pkg-yum@lists.postgresql.org; Wed, 04 Nov 2026 11:13:38 +0000 ARC-Authentication-Results: i=1; ma04.lansolnet.com 1; spf=pass reason=mailfrom (ip=194.126.198.214, headerfrom=nidsa.net) smtp.mailfrom=nidsa.net smtp.helo=mail.cloud4partner.com; ARC-Message-Signature: a=rsa-sha256; bh=Oo3383nZJxAQ3oVvcwqTgYkUVaUg88wuXknfSO4MiRE=; c=relaxed/relaxed; d=hornetsecurity.com; h=from:to:date:subject:mime-version:; i=1; s=hse1; t=1793790814; b=jzo6kCvY8ydpnMiH94YWFi9KQF6QKLfkFS1V/Yy60hZ5zTN/CBZuWSj/9EBLE0o5FRquxnYE ADAtbTWUlAyaJPQQoDwPV9ixZx7CPBxm15sad/NRfYNV0XmO+/IirV2nZ8K1SmJxkSrcvsC+iv1 drh6pW3DkV5lErqGs5a64fXJ2VYoXQp8vPZCuWT+b1IKhRaJ3Z1f+HUH32KcDMBTSeKygacqt+7 59ZaRE9iZlKKlRZdNMCjq7taG8j1dJgAmNV0uChcaQuzsHKBsGPhsIrVSu7iOINFrIZFxDanrMZ L9mH95ntIhpKV626vc7ZfoQbILG3DPmCfQI6b4k5AInLA== ARC-Seal: a=rsa-sha256; cv=none; d=hornetsecurity.com; i=1; s=hse1; t=1793790814; b=faAPryP5Fl/w/BHUfw+XSaHRSlbp/xwjjjDcBeexg916CTWdHQPpPlxUsb+LDX819EbGgTz6 MW7qlztloHlk/d5zEDzd2l/gLXB7Crr1nm3WEYk4Z67c8a2NVzy/nsg2Up6p3bOrlBF71fMbHIG mmQybAp+kyDXFUHRVrzlRTYg5bBkMWds0hxam8vZosyqcLc5t4inkjBj9HeFPi39ziQaVPFSBNT 70exA/5qL/JYJH9sCli/edDNyt7cpG0MmpPLOUP+fgqYC+b9pe6V8n7T+Q34RVYMIvJ/eYoARjl 7504QF/SZ4W5vSH07SqM++eTfXGN1K3nDELAAhE7nNuAw== Received: from unknown ([194.126.198.214]) by ma04; Wed, 04 Nov 2026 12:13:34 +0100 Received: from MBX200E.cloud4partner.com (192.168.8.234) by MBX200F.cloud4partner.com (192.168.8.235) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA384_P521) id 15.1.2507.44; Wed, 4 Nov 2026 12:13:31 +0100 Received: from MBX200E.cloud4partner.com ([fe80::65e4:9dc9:128b:185d]) by MBX200E.cloud4partner.com ([fe80::65e4:9dc9:128b:185d%15]) with mapi id 15.01.2507.044; Wed, 4 Nov 2026 12:13:31 +0100 From: Hans Buschmann To: "devrim@gunduz.org" , "pgsql-pkg-yum@lists.postgresql.org" Subject: Re: Fedora 45 release timing Thread-Topic: Fedora 45 release timing Thread-Index: AQHdbENgZm1HT/3mgEqY4qSMBcLgzQ== Date: Wed, 4 Nov 2026 11:13:31 +0000 Message-ID: <9998a87378fa43ce9102b3eac3be387c@nidsa.net> Accept-Language: de-DE, en-US Content-Language: de-DE X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [192.168.8.240] Content-Type: multipart/alternative; boundary="_000_9998a87378fa43ce9102b3eac3be387cnidsanet_" MIME-Version: 1.0 X-cloud-security-sender: buschmann@nidsa.net X-cloud-security-recipient: pgsql-pkg-yum@lists.postgresql.org X-cloud-security-Virusscan: CLEAN X-cloud-security-disclaimer: This E-Mail was scanned by E-Mailservice on ma04 with 041A5B817A3 X-cloud-security-connect: unknown[194.126.198.214], TLS=1, IP=194.126.198.214 X-cloud-security: scantime:.7222 List-Id: List-Help: List-Subscribe: List-Post: List-Owner: List-Archive: Archived-At: Precedence: bulk --_000_9998a87378fa43ce9102b3eac3be387cnidsanet_ Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hello Devrim, Given the current delay of Postgres 19 and the ongoing intense bug hunting = caused by LLM AI activity I can understand the delay of more than a week. But the experience of real world users on development and production should= be much smoother then that: I use Fedora Server for production for about 10 years, mostly on virtual ma= chines and some bare metals. In my experience running postgres on fedora server never causes trouble (li= ke no installation, missing libraries). Even manual compiling with meson is= very smooth, some newer compililer warnings of the leading edge GCC compil= er where reported and quickly fixed. I always test and self-compile the fre= shest beta1 of postgres to ensure the compatibility with our application. This all leads me to the main point: the repositories for a new fedora rele= ase should be provided when fedora beta is released !!! -- this is well planned in advance -- it has a freeze of new features, so the tooling should be identical/near= the same as for general release -- this gives the opportunity to test postgres on the new release -- this eliminates the work pressure on fedora GA release date (which may s= lip) -- after fedora GA the beta installation is easily converted to the final v= ersion by dnf upgrade -- the availability should be restricted to official releases, no beta/rcs,= perhaps by skipping the oldest release soon going EOL This practice is followed for a long time by remis's RPM repository concern= ing php: -- you have early access to beta/rc versions of the new php, also on the up= comming fedora release. The work to be done should be no different to the after-GA-providing, only = 6 to 8 weeks earlier. Tools should work flawlessly. The problems arise when you upgrade fedora without any PGDG RPMs for postgr= es available (here 44 to 45): -- Error messages error 404 of RPMs not found -- automatic install/update from the PGDG Postgres installation to the fedo= ra internal installation packages, which is easily overseen between about 1= 000++ update packages -- perhaps version regression because of fedora not providing the newest mi= nor version. -- different installation paths, logging, configuration, setup procedure et= c. -- manual removal (of fedora provided packages) triggers often the removing= of other required packages (like perl), which must be reinstalled later on -- users visiting the website don't find a corresponding RPM version, albei= t Fedora support is claimed. On the postgres website it can be marked that the support on a beta version= of the operating system like fedora (or RHEL, debian etc) is only prelimin= ary until GA is available. The early availability ensures every user that his tooling for the new oper= ating system is already available (like the important external tools postgr= es, php etc) so that the OS upgrade can be planned smoothely!! The same arguments ar valid for the other operating systems... One last remark: In time of a minor release of postgres (normally at the scheduled dates) AL= L parts of the new code should be provided, especially the libpq library. I= n the past libpq arrived more then a wek later, so new uncertainities of ad= ministration, supervising, pending security patches etc. Thank you Hans Buschmann --_000_9998a87378fa43ce9102b3eac3be387cnidsanet_ Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

Hello Devrim,


Given the current delay of Postgres 19 and the ongoing intense bug hunti= ng caused by LLM AI activity I can understand the delay of more than a week= .


But the experience of real world users on development and production sho= uld be much smoother then that:



I use Fedora Server for production for about 10 years, mostly on vi= rtual machines and some bare metals.


In my experience running postgres on fedora server never causes trouble = (like no installation, missing libraries). Even manual compiling with meson= is very smooth, some newer compililer warnings of the leading edge GCC&nbs= p;compiler where reported and quickly fixed. I always test and self-compile the freshest beta1 of postgres to en= sure the compatibility with our application.


This all leads me to the main point: the repositories for a new fedora r= elease should be provided when fedora beta is released !!!


-- this is well planned in advance

-- it has a freeze of new features, so the tooling should be identical/n= ear the same as for general release

-- this gives the opportunity to test postgres on the new release

-- this eliminates the work pressure on fedora GA release date (whi= ch may slip)

-- after fedora GA the beta installation is easily converted to the fina= l version by dnf upgrade

-- the availability should be restricted to official releases, no beta/r= cs, perhaps by skipping the oldest release soon going EOL


This practice is followed for a long time by remis's RPM repository conc= erning php:

-- you have early access to beta/rc versions of the new php, also on the= upcomming fedora release.


The work to be done should be no different to the after-GA-providing, on= ly 6 to 8 weeks earlier. Tools should work flawlessly.


The problems arise when you upgrade fedora without any PGDG RPMs for pos= tgres available (here 44 to 45):


-- Error messages error 404 of RPMs not found

-- automatic install/update from the PGDG Postgres installation to the f= edora internal installation packages, which is easily overseen between abou= t 1000++ update packages

-- perhaps version regression because of fedora not providing the newest= minor version.

-- different installation paths, logging, configuration, setup proc= edure etc.

-- manual removal (of fedora provided packages) triggers often the remov= ing of other required packages (like perl), which must be reinstalled later= on

-- users visiting the website don't find a corresponding RPM version, al= beit Fedora support is claimed.


On the postgres website it can be marked that the support on a beta vers= ion of the operating system like fedora (or RHEL, debian etc) is only preli= minary until GA is available.


The early availability ensures every user that his tooling for the new o= perating system is already available (like the important external tool= s postgres, php etc) so that the OS upgrade can be planned smoothely!!


The same arguments ar valid for the other operating systems...



One last remark:


In time of a minor release of postgres (normally at the scheduled dates)= ALL parts of the new code should be provided, especially the libpq li= brary. In the past libpq arrived more then a wek later, so new uncertainiti= es of administration, supervising, pending security patches etc.



Thank you


Hans Buschmann










--_000_9998a87378fa43ce9102b3eac3be387cnidsanet_--