devel-announce October 2023

devel-announce@lists.fedoraproject.org

3 participants
12 discussions

Fedora Linux 39 Final is NO-GO (#3)
by Adam Williamson 26 Oct '23

26 Oct '23

Due to outstanding blocker bugs[1], F39 Final RC-1.2 was declared NO-GO in today's Go/No-GO meeting[2]. The next Fedora Linux 39 Final Go/No-Go meeting[3] will be held at 1700 UTC on Thursday 2 November in #fedora-meeting. We will aim for the "target date #3" milestone of 7 November. The release schedule[4] has been updated accordingly. [1] https://qa.fedoraproject.org/blockerbugs/milestone/39/final/buglist [2] https://meetbot-raw.fedoraproject.org/fedora-meeting/2023-10-26/f39-final-g… [3] https://calendar.fedoraproject.org/meeting/10627/?from_date=2023-10-30 [4] https://fedorapeople.org/groups/schedule/f-39/f-39-key-tasks.html -- Adam Williamson (he/him/his) Fedora QA Fedora Chat: @adamwill:fedora.im | Mastodon: @adamw@fosstodon.org https://www.happyassassin.net

1 0

Fedora Linux 39 Final Go/No-Go meeting tomorrow (Thursday)
by Adam Williamson 25 Oct '23

25 Oct '23

The Fedora Linux 39 Final Go/No-Go[1] meeting is scheduled for Thursday 26 October at 1700 UTC in #fedora-meeting. At this time, we will determine the status of the F39 Final for the 31 October target date[2]. For more information about the Go/No-Go meeting, see the wiki[3]. [1] https://calendar.fedoraproject.org/meeting/10627/ [2] https://fedorapeople.org/groups/schedule/f-39/f-39-key-tasks.html [3] https://fedoraproject.org/wiki/Go_No_Go_Meeting -- Adam Williamson (he/him/his) Fedora QA Fedora Chat: @adamwill:fedora.im | Mastodon: @adamw@fosstodon.org https://www.happyassassin.net

1 0

Fedora Linux 39 Final is NO-GO (#2)
by Adam Williamson 19 Oct '23

19 Oct '23

Due to outstanding blocker bugs[1], we do not have a release candidate for Fedora Linux 39 Final. Today's Go/No-Go meeting is cancelled. The next Fedora Linux 39 Final Go/No-Go meeting[2] will be held at 1700 UTC on Thursday 26 October in #fedora-meeting-1. We will aim for the "target date #2" milestone of 31 October. The release schedule[3] will be updated accordingly soon. [1] https://qa.fedoraproject.org/blockerbugs/milestone/39/final/buglist [2] https://calendar.fedoraproject.org/meeting/10627 [2] https://fedorapeople.org/groups/schedule/f-39/f-39-key-tasks.html -- Adam Williamson (he/him/his) Fedora QA Fedora Chat: @adamwill:fedora.im | Mastodon: @adamw@fosstodon.org https://www.happyassassin.net

1 0

F40 Change Proposal: Removing OpenSSL 1.1 package (System-Wide)
by Aoife Moloney 17 Oct '23

17 Oct '23

Wiki Link: https://fedoraproject.org/wiki/Changes/RemoveOpensslCompat Discussion.fpo Link: https://discussion.fedoraproject.org/t/f40-change-proposal-removing-openssl… == Summary == We are going to remove the openssl1.1 package from Fedora 40. == Owner == * Name: [[User:DmitryBelyavskiy| Dmitry Belyavskiy]] * Email: dbelyavs(a)redhat.com == Detailed Description == In Fedora 36 we switched to OpenSSL 3.0 branch. This is a brand new version with new architecture. We left the openssl1.1 package for the applications that were unable to switch to the new API/architecture, 3rd-party applications, etc. The package was marked as deprecated in F37. OpenSSL 1.1.1 has reached EOL in September 2023. We want to remove it from Fedora. == Feedback == == Benefit to Fedora == This proposal ensures than no new packages in Fedora will use the deprecated OpenSSL version that will cause an overall increase of security/stability. It will also reduce the maintenance burden for the OpenSSL maintainers, especially when new CVEs are published. == Scope == * Proposal owners: provide assistance in migration to other developers. * Other developers: Patch their packages to work with OpenSSL 3.0. * Release engineering: This feature doesn't require coordination with release engineering. * Policies and guidelines: N/A (not needed for this Change)  * Trademark approval: N/A (not needed for this Change) * Alignment with Community Initiatives: == Upgrade/compatibility impact == 3rd-party packages depending on OpenSSL 1.1.1 should be replaced with new versions using new OpenSSL 3.0+. == How To Test == OpenSSL 1.1 should not be available to install from Fedora repository. No packages should depend on OpenSSL 1.1.1. == User Experience == Shouldn't be affected. == Dependencies == We have found at least the following packages depending on OpenSSL 1.1: * gloo-0.5.0^git20230824.01a0c81-6.fc40.src.rpm * opensmtpd-6.8.0p2-12.fc39.src.rpm * python3.6-3.6.15-20.fc39.src.rpm == Contingency Plan == None. * Contingency mechanism: (What to do? Who will do it?) Package owners should update their packages to remove the dependency * Contingency deadline: beta freeze * Blocks release? Yes == Documentation == Should be mentioned in Release Notes. == Release Notes == openssl1.1 package is removed and should not be used by any packages. -- Aoife Moloney Product Owner Community Platform Engineering Team Red Hat EMEA Communications House Cork Road Waterford

1 0

F41 Change Proposal: Python3.13 (System-Wide)
by Aoife Moloney 17 Oct '23

17 Oct '23

Wiki Link: https://fedoraproject.org/wiki/Changes/Python3.13 Discussion.fpo Link: https://discussion.fedoraproject.org/t/f41-change-proposal-python-3-13-syst… == Summary == Update the Python stack in Fedora from Python 3.12 to Python 3.13, the newest major release of the Python programming language. == Owner == * Name: [[User:Churchyard|Miro Hrončok]] * Name: [[User:Ksurma|Karolina Surma]] * Email: python-maint(a)redhat.com == Detailed Description == We would like to upgrade Python to 3.13 in Fedora 41 thus we are proposing this plan early. See the upstream notes at [https://docs.python.org/dev/whatsnew/3.13.html#what-s-new-in-python-3-13 What's new in 3.13]. === Important dates and plan === * 2023-05-22: Python 3.13 development begins * 2023-10-13: Python 3.13.0 alpha 1 ** Package it as {{package|python3.13}} for testing purposes ** Start the bootstrap procedure in Copr ** Do a mass rebuild against every future release in Copr * 2023-11-21: Python 3.13.0 alpha 2 * 2023-12-19: Python 3.13.0 alpha 3 * 2024-01-16: Python 3.13.0 alpha 4 * 2024-02-06: Branch Fedora 40, Rawhide becomes future Fedora 41 ** The earliest point when we can start rebuilding in Koji side-tag * 2024-02-13: Python 3.13.0 alpha 5 * 2024-03-12: Python 3.13.0 alpha 6 * 2024-04-09: Python 3.13.0 alpha 7 * 2024-05-07: Python 3.13.0 beta 1 ** No new features beyond this point * 2024-05-28: Python 3.13.0 beta 2 ** The ideal point when we can start rebuilding in Koji * 2024-06-04: Expected side tag-merge (optimistic) * 2024-06-18: Python 3.13.0 beta 3 * 2024-06-25: Expected side tag-merge (realistic) * 2024-07-15: Expected side tag-merge (pessimistic) * 2024-07-16: Python 3.13.0 beta 4 * 2024-07-17: Fedora 41 Mass Rebuild ** The mass rebuild happens with the fourth or third beta. We might need to rebuild Python packages later in exceptional case. ** If the Koji side-tag is not merged yet at this point, we defer the change to Fedora 42. * 2024-07-30: Python 3.13.0 candidate 1 ** This serves as "final" for our purposes. * 2024-08-06: Branch Fedora 41, Rawhide becomes future Fedora 42 * 2024-08-06: Fedora 41 Change Checkpoint: Completion deadline (testable) * 2024-08-20: Fedora Beta Freeze ** If rebuild with 3.13.0rc1 is needed, we should strive to do it before the freeze - there is a window of 3 weeks. * 2024-09-03: Python 3.13.0 candidate 2 * 2024-09-10: Fedora 41 Beta Release (Preferred Target) ** Beta will likely be released with 3.13.0rc1. * 2024-09-17: Fedora 41 Beta Target date #1 * 2024-10-01: Python 3.13.0 final * 2024-10-01: Fedora 39 Final Freeze ** We'll update to 3.13.0 final using a freeze exception. * 2024-10-15: Fedora 39 Preferred Final Target date * 2024-10-22: Fedora 39 Final Target date #1 (From [https://peps.python.org/pep-0719/#schedule Python 3.13 Release Schedule] and [https://fedorapeople.org/groups/schedule/f-41/f-41-key-tasks.html Fedora 41 Release Schedule].) The schedule might appear somewhat tight for Fedora 41, but Python's [https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org… annual release cycle was adapted for Fedora] and this worked fine since Python 3.9 and Fedora 33. It is now common that Python is upgraded on a similar schedule in every odd-numbered Fedora release. Note that upstream's "release candidates" are frozen except for blocker bugs. Since we can and will backport blocker fixes between Fedora and upstream, we essentially treat the Release Candidate as the final release. == Benefit to Fedora == Fedora aims to showcase the latest in free and open-source software - we should have the most recent release of Python 3. Packages in Fedora can use the new features from 3.13. There's also a benefit to the larger Python ecosystem: by building Fedora's packages against 3.13 while it's still in development, we can catch critical bugs before the final 3.13.0 release. == Scope == We will coordinate the work in a side tag and merge when ready. * Proposal owners: *# Introduce {{package|python3.13}} for all Fedoras *# Prepare stuff in Copr as explained in description. *# Update {{package|python-rpm-macros}} so {{package|python3.13}} builds {{package|python3}} *# Build {{package|python3.13}} as the main Python *# Mass rebuild all the packages that runtime require `python(abi) = 3.12` and/or `libpython3.12.so.1.0` (~4000 known packages in October 2023) *# Build {{package|python3.13}} as a non-main Python * Other developers: Maintainers of packages that fail to rebuild during the rebuilds will be asked, using e-mail and bugzilla, to fix or remove their packages from the distribution. If any issues appear, they should be solvable either by communicating with the respective upstreams first and/or applying downstream patches. Also, the package maintainers should have a look at: [https://docs.python.org/dev/whatsnew/3.13.html#porting-to-python-3-13 Porting to Python 3.13]. The python-maint team will be available to help with fixing issues. * Release engineering: [https://pagure.io/releng/issue/11728 #11728] * Policies and guidelines: N/A (not needed for this Change) * Trademark approval: N/A (not needed for this Change) * Alignment with Community Initiatives: N/A == Upgrade/compatibility impact == All the packages that depend on Python 3 must be rebuilt. User written Python 3 scripts/applications may require a small amount of porting, but mostly Python 3.12 is forward compatible with Python 3.13. == How To Test == Interested testers do not need special hardware. If you have a favourite Python 3 script, module, or application, please test it with Python 3.13 and verify that it still works as you would expect. If the application you are testing does not require any other modules, you can test it using {{package|python3.13}} even before this change is implemented, in Fedora 37, 38, 39 or 40. In case your application requires other modules, or if you are testing an rpm package, it is necessary to install the 3.13 version of the python3 rpm. Right now that rpm is available in copr, along with all other python packages that build successfully with python 3.13. See https://copr.fedorainfracloud.org/coprs/g/python/python3.13/ for detailed instructions on how to enable Python 3.13 copr for mock. Once the change is in place, test if your favorite Python apps are working as they were before. File bugs if they don't. == User Experience == Regular distro users shouldn't notice any change in system behaviour other than the Python 3 interpreter will be in version 3.13. == Dependencies == 4400+ packages depend on Python 3 and ~4000 packages need rebuilding when Python is upgraded. See scope section. == Contingency Plan == * Blocks release? Yes, we'd like to block Fedora 41 release on at least 3.13.0rc1 == Documentation == [https://peps.python.org/pep-0719/ Python 3.13 Release Schedule] [https://docs.python.org/dev/whatsnew/3.13.html#what-s-new-in-python-3-13 What's new in 3.13] [https://docs.python.org/dev/whatsnew/3.13.html#porting-to-python-3-13 Porting to Python 3.13] == Release Notes == -- Aoife Moloney Product Owner Community Platform Engineering Team Red Hat EMEA Communications House Cork Road Waterford

1 0

F40 Change Proposal: Update to Pydantic Version 2 (Self-Contained)
by Aoife Moloney 17 Oct '23

17 Oct '23

Wiki Link: https://fedoraproject.org/wiki/Changes/Update_To_Pydantic_Version_2 Discussion.fpo Link: https://discussion.fedoraproject.org/t/f40-change-proposal-update-to-pydant… This document represents a proposed Change. As part of the Changes process, proposals are publicly announced in order to receive community feedback. This proposal will only be implemented if approved by the Fedora Engineering Steering Committee. == Summary == {{package|python-pydantic}}, a Python data validation library, will be updated from 1.10.z to 2.y.z. The Change owners will perform a test rebuild and work with package maintainers and upstreams to port code. == Owner == * Name: Maxwell G; Benjamin Beasley; Python SIG * Email: maxwell(a)gtmx.me; code(a)musicinmybrain.net; python-devel(a)lists.fedoraproject.org == Detailed Description == Pydantic is a Python data validation library. Recently, upstream released major version 2 with many fixes and performance improvements but also multiple breaking API changes. The new release relies on {{package|python-pydantic-core}}, which provides the core validation and serialization functionality used by pydantic v2. pydantic-core is a Rust PyO3 Python extension module. We will update pydantic to version 2 after making sure dependent packages are accounted for. === Porting strategies === Pydantic v2 contains some breaking API changes that may require porting if packages depend on old functionality. Upstream provides a detailed [https://docs.pydantic.dev/latest/migration/ Migration Guide] that describes the breaking API changes. Projects that are not ready to port to pydantic v2 may use the <code>pydantic.v1</code> compatibility module which contains a full copy of the old pydantic v1 code. https://github.com/matrix-org/synapse/pull/16332 exemplifies this approach. As shown in the above PR, projects can use conditional imports to import <code>pydantic.v1</code> if pydantic v2 is in use and otherwise fall back to importing <code>pydantic</code> if pydantic v1 is in use. Maintaining compatibility for both pydantic v1 and v2 in the same codebase without the compatibility module is possible but may not be practical for larger, more complex usecases. fedrq [https://git.sr.ht/~gotmax23/fedrq/commit/35d64cd2a8bd0ea8210d8b086e14e3d666… uses this approach] and silences <code>pydantic.PydanticDeprecatedSince20</code> warnings. Many of these deprecation warnings cannot be resolved without dropping support for pydantic v1. == Feedback == I did not receive much feedback as of yet, but I expect this change to be uncontroversial. == Benefit to Fedora == Fedora will have the latest version of pydantic in its repositories. The new version touts a significant performance boost, amongst other improvements. == Scope == * Proposal owners: ** Preform an [https://hackmd.io/@python-maint/rJSm5WC9Y impact check] ** Make an inventory of failures or packages that have a hard dependency on pydantic v1. ** Submit distgit PRs and/or file upstream issues to fix incompatibilities in dependent packages ** Submit new python-pydantic-settings and python-pydantic-extra-types packages for review ** Wait two weeks for packagers to review PRs. Those that remain compatible with pydantic v1 can be merged immediately. ** Build pydantic v2 package and any outstanding PRs in a side tag. * Other developers: ** Test your package to ensure it still builds and functions with pydantic v2 ** Help with package reviews ** Review distgit PRs to fix incompatibilities == Upgrade/compatibility impact == Pydantic version 2 has some breaking API changes. See upstream's [https://docs.pydantic.dev/latest/migration/ Migration Guide]. == How To Test == You can perform test builds like this: <pre> $ copr mock-config gotmax23/pydanticv2-testing fedora-rawhide-x86_64 > ~/.config/pydanticv2.cfg $ fedpkg --release=rawhide mockbuild --root pydanticv2 </pre> == User Experience == This isn't a particularly user visible change. Users of Python applications that utilize Pydantic should notice a performance improvement. == Dependencies == See https://gtmx.me/Wiki/Fedora/pydantic-v2-update/#lists for a list of dependent packages. The Change owners will work with these packages' maintainers to ensure the packages remain functional with the new pydantic version. == Contingency Plan == * Contingency mechanism: N/A (not a System Wide Change) * Contingency deadline: N/A (not a System Wide Change) * Blocks release? N/A (not a System Wide Change) == Documentation == See https://gtmx.me/Wiki/Fedora/pydantic-v2-update/ for the current status and notes. == Release Notes == python3-pydantic was updated from 1.10.z to 2.y.z. Pydantic v2 brings performance improvements and an API refactoring, amongst other changes. See upstream's [https://docs.pydantic.dev/latest/migration/ Migration Guide] for a full inventory of breaking changes. -- Aoife Moloney Product Owner Community Platform Engineering Team Red Hat EMEA Communications House Cork Road Waterford

1 0

F40 proposal: Fedora Atomic Desktops (Self-Contained Change proposal)
by Adam Williamson 17 Oct '23

17 Oct '23

https://fedoraproject.org/wiki/Changes/AtomicDesktops This document represents a proposed Change. As part of the Changes process, proposals are publicly announced in order to receive community feedback. This proposal will only be implemented if approved by the Fedora Engineering Steering Committee. = Fedora Atomic Desktops = == Summary == We will regroup all desktop, rpm-ostree based variants of Fedora under the Fedora Atomic Desktops name. Each individual variant (Silverblue, Kinoite, Sericea, Onyx) may keep their name as is. While this is a Change Request, it is not addressed at FESCo but at the Fedora Council as this is not a technical change but a marketing / policy one. == Owner == * Name: Timothée Ravier * Email: siosm(a)fedoraproject.org * Name: Micah Abbott * Email: miabbott(a)redhat.com * Name: Fabio Alessandro * Email: fale(a)fedoraproject.org * Name: Joshua Strobl * Email: joshua(a)buddiesofbudgie.org == Detailed Description == The [https://fedoraproject.org/ Fedora website] currently uses the term "Immutable Desktops" to regroup all desktop, `rpm-ostree` based Fedora variants. The term "immutable" is confusing to users, has been the source of many confusion and does not accurately reflect the advantages of those variants. Thus we want to regroup those variants under a "new" name. The advantage of the Atomic name is that it already was a brand in the past, and it’s both technically accurate, short, but non-descriptive and non-restrictive so that people would have pre-conceived ideas like with “immutable”. It is important to state that this change does not intend to revive [https://www.projectatomic.io Project Atomic] which acted as an umbrella for Atomic Host, Silverblue, and container technologies. This change is merely borrowing the "Atomic" name as a way to better describe the `rpm-ostree` based desktop operating systems. We've created the [https://fedoraproject.org/wiki/SIGs/AtomicDesktops Fedora Atomic Desktops SIG] to co-ordinate across-variants work. One of the long term goals of this SIG is to work on making those variants the default option in Fedora, thus removing the need for a distinct name. We will not require existing variants (Silverblue, Kinoite, Sericea, Onyx) to be renamed, as those brands already have some traction and history. It will be up to the decision of the SIGs if they want to rename the existing variants as a result of this change. We will not rename the ostree refs or do any technical changes related to this change to avoid costly and mostly useless or invisible work. New variants will use the Fedora <Desktop> Atomic name. For example, in the case of XFCE it would be: Fedora XFCE Atomic. Note that both Fedora CoreOS and Fedora IoT will remain as is and will not be regrouped under this brand, which only focuses on desktops. Both CoreOS & IoT variants have a strong brand on their own and do not suffer from the immutable naming. == Feedback == A lot of other names have been suggested ("package mode", "image mode", "reprovisionable", "anti-hysteresis") but so far those are either "too technical", "too long" or "too ambiguous/imprecise". Another suggested option was to rename all variants under the Silverblue name (Silverblue GNOME, Kinoite -> Silverblue KDE, etc.). This however creates too long names and would likely create confusion as user would expect all systems to share a Silverblue base, which is not the case. References: * https://blog.verbum.org/2020/08/22/immutable-%e2%86%92-reprovisionable-anti… * https://theevilskeleton.gitlab.io/2023/08/29/misconceptions-about-immutable… * https://gitlab.com/fedora/websites-apps/fedora-websites/fedora-websites-3.0… * https://fedoraproject.org/wiki/Changes/Silverblue * https://discussion.fedoraproject.org/t/i-am-ok-with-fedora-atomic-brand-now… * https://discussion.fedoraproject.org/t/creating-the-fedora-atomic-desktops-… * https://fedoraproject.org/wiki/SIGs/AtomicDesktops == Benefit to Fedora == The main benefits are: less confusion for users / podcasters / reviewers, better branding / marketing. == Scope == * Proposal owners: Submit / review pull-requests to update the name on the Fedora Website * Other developers: N/A * Release engineering: N/A * Policies and guidelines: N/A * Trademark approval: [https://pagure.io/Fedora-Council/tickets/issue/361 #361] * Alignment with Community Initiatives: N/A == Upgrade/compatibility impact == No technical changes so not impact on users. == How To Test == Nothing to test beyond the website changes. == User Experience == This should not be noticeable by users from a technical perspective. == Dependencies == N/A == Contingency Plan == * Contingency mechanism: Everything stays as is. * Contingency deadline: N/A (not a System Wide Change) * Blocks release? No == Documentation == We'll try to update the documentation and figure out a way to share more of the content between those variants. == Release Notes == To be done. -- Adam Williamson (he/him/his) Fedora QA Fedora Chat: @adamwill:fedora.im | Mastodon: @adamw@fosstodon.org https://www.happyassassin.net

1 1

F40 Change Proposal: PostgreSQL 16 (Self-Contained)
by Aoife Moloney 16 Oct '23

16 Oct '23

Wiki Link: https://fedoraproject.org/wiki/Changes/PostgreSQL_16 Discussion.fpo Link: https://discussion.fedoraproject.org/t/f40-change-proposal-postgresql-16-se… This is a proposed Change for Fedora Linux. This document represents a proposed Change. As part of the Changes process, proposals are publicly announced in order to receive community feedback. This proposal will only be implemented if approved by the Fedora Engineering Steering Committee. = PostgreSQL 16 = {{Change_Proposal_Banner}} == Summary == Update of default PostgreSQL stream (`postgresql` and `libpq` components) in Fedora from version 15 to version 16. Since modularity was marked as retired, there will also be a change in the packaging concept. == Owner == * Name: [[User:fjanus| Filip Januš]] * Email: fjanus(a)redhat.com == Detailed Description == The original package `PostgreSQL` will be retired. Since Fedora 40 there will be a separate component in the rpms namespace for each shipped PostgreSQL. So postgresql15 for PG15 and postgresql16 for PG16. Each PG stream will provide `postgresql` symbol and conflict with each other. The choice of default stream will be made by building sub-packages with appropriate names. (using `-n` in the `files` section). This approach should solve, among other things, the problem of possible duplicity. There is the following problem if we kept one version to be shipped in unversioned postgresql component: F40 - there is currently a default version PG15 (postgresql -> 15) F40 - there will be added version PG16 (postgresql -> 15, postgresql16) F40 - switch default version to PG16 (postgresql -> 16, postgresql16, new postgresql15) If PG15 is to be present in F40, to follow non-modular concept the new package is necessary. BUt such switch leads to duplicit packages postgresql and postgresql16 with the same content. Using just versioning packages would solve the described issue. The proposed solution is described below. Plan of the future structure of Postgresql versioning: Fedora 40 where PostgreSQL 16 is default: postgresql SRPM -> deprecated postrgresql15 SRPM -> postgresql15, postgresql15-server, ... postrgresql16 SRPM -> postgresql, postgresql-server, ... postrgresql17 SRPM -> postgresql17, postgresql17-server, ... Fedora 42 where PostgreSQL 17 is default: postgresql SRPM -> deprecated postrgresql15 SRPM -> postgresql15, postgresql15-server, ... postrgresql16 SRPM -> postgresql16, postgresql16-server, ... postrgresql17 SRPM -> postgresql, postgresql-server, ... This also involves updating and rebuilding the PostgreSQL plugins that depend on postgresql server. == Feedback == == Benefit to Fedora == The latest stable software is provided for Fedora users. Modules are no longer needed. == Scope == * Proposal owners: **Prepare PostgreSQL 16 as the default stream **Prepare PostgreSQL 15 as a non-default stream **Check software that requires or depends on `postgresql-server` or `libpq` packages for incompatibilities **Build PostgreSQL 16 (postgresql and libpq) for Rawhide **Build PostgreSQL 15 for Rawhide **Rebuild dependent packages against PostgreSQL 16 * Other developers: * Release engineering: [https://pagure.io/releng/issues #Releng issue number] * Policies and guidelines: N/A (not needed for this Change) * Trademark approval: N/A (not needed for this Change) * Alignment with Community Initiatives: == Upgrade/compatibility impact == The PostgreSQL client library (libpq component) is compatible. So, there shouldn't be any compatibility issues, but rebuild of the dependent components is recommended. Server plugins might require a newer version update because they sometimes have explicit server requirements. == How To Test == All PG server plugins should be installable. `postgresql-setup --upgrade` command should succeed. Test that all other software runs well with PostgreSQL 16. == User Experience == The users will have to upgrade their databases the same way as between major PostgreSQL versions, aka `postgresql-setup --upgrade` after installing PostgreSQL 16 server packages. If users want to stick with PostgreSQL 15 for a little longer, there will be PostgreSQL 15 as nondefault PostgreSQL stream == Dependencies == There are some packages (mostly server plugins), that build on top of PostgreSQL. Since the separation of PostgreSQL client library (libpq component), only packages that build server plugins should use postgresql package in BuildRequires. Others should use libpq. In the case of Postgresql-server, a rebuild should be done to ensure all potential binary incompatibilities are handled. * PostgreSQL server dependecies ** perl-DBD-Pg ** pgaudit ** qt ** qt3 ** qt5-qtbase ** postgres-decoderbufs ** gambas3 ** kdb ** kea ** libpqxx ** openvas-manager ** orafce ** pg-semver ** pgRouting ** pgadmin3 ** pgsphere ** postgis ** postgresql-ip4r ** postgresql-pgpool-II ** qt3 ** rdkit ** rhdb-utils ** timescaledb ** pg_repack == Contingency Plan == * Contingency mechanism: (What to do? Who will do it?) N/A (not a System Wide Change) * Contingency deadline: N/A (not a System Wide Change) * Blocks release? N/A (not a System Wide Change), Yes/No Revert the changes and provide PostgreSQL 15 only. == Documentation == Upgrade strategy: https://www.postgresql.org/docs/16/upgrading.html N/A (not a System Wide Change) == Release Notes == Release notes for PostgreSQL 16 release: https://www.postgresql.org/docs/16/index.html Overall overview of the changes and improvements: https://www.postgresql.org/docs/16/release-16.html -- Aoife Moloney Product Owner Community Platform Engineering Team Red Hat EMEA Communications House Cork Road Waterford

1 0

F40 Change Proposal: F40 MariaDB MySQL repackaging (Self-Contained)
by Aoife Moloney 16 Oct '23

16 Oct '23

Wiki Link: https://fedoraproject.org/wiki/Changes/F40_MariaDB_MySQL_repackaging Discussion.fpo Link: https://discussion.fedoraproject.org/t/f40-change-proposal-f40-mariadb-mysq… This is a proposed Change for Fedora Linux. This document represents a proposed Change. As part of the Changes process, proposals are publicly announced in order to receive community feedback. This proposal will only be implemented if approved by the Fedora Engineering Steering Committee. = F40 MariaDB & MySQL repackaging = [[Category:Package MariaDB]] {{Change_Proposal_Banner}} == Summary == A bigger set of smaller changes which I want to extend visibility for: * Drop builds for i686 architecture * Rename package 'community-mysql' to 'mysql' and Stop providing 'mysql' symbols by package 'mariadb' * Drop cross-installation functionality * Switch to the versioned layout of MariaDB and MySQL packages * Introduce MariaDB 10.11 and MySQL 8.1 * Change the default MariaDB version in Fedora from 10.5 to 10.11 == Owner == * Name: [[User:mschorm| Michal Schorm]] * Email: mschorm(a)redhat.com == Detailed Description == Most (if not all) of the changes I propose here doesn't probably need such a formal process as Fedora Change is. What I want is primarily to increase the visibility of the changes, and preserve those information in a permanent document, allowing both users and maintainers to find this page both now, and in the future. The changes described here are something I mostly need to be done, as a maintainer. They are my informed decisions backed up by the expertise I've gained over the many years I maintain these packages. I believe in the community discussion to focus primarily the technical aspects of the changes to help me to implement them correctly, bug-free, rather than whether to apply these changes at all. Moreover, the changes here are described in a specific order. The order is based on how each task logically follow other. In case some of the changes would be found problematic, it might be dropped from the proposal (with all changes that directly relies on it) '''Drop builds for i686 architecture:''' Both MariaDB and MySQL databases are huge pieces of software, taking hours to compile and test, taking up noticeable amount of system resources on our builders and any attached system (e.g. repositories). At the same time, I can hardly imagine anyone running those databases in production on i686 nowadays. Nothing, expect the 'mariadb' and 'community-mysql' packages, which provide the DB servers and client application. Drop of the other packages from the stack is not part of this proposal. So for now, packages like 'mariadb-connector-c' (which provides the client library 'libmariadb.so') and the ODBC or JAVA connectors will keep to build and ship their i686 variants. However the DB servers themselves are IMO expendable. On top of the resources saved, I also keep finding bugs and issues, that are exclusive for the i686 architecture. They are more often than not in the tooling around, rather than the package itself, but it keeps needlessly devouring my time and energy anyway. Last but not least, all maintainers are currently encouraged to drop i686 support, starting from leaf packages. https://fedoraproject.org/wiki/Changes/EncourageI686LeafRemoval So I have to make sure the MariaDB and MySQL are leaf package on i686 architecture, so I can drop their i686 variants too. I've started the process by fixing https://bugzilla.redhat.com/show_bug.cgi?id=2241091 I use the following code to find out which packages requires which provides: #!/bin/bash mkdir -p RESULTS for REPO in "rawhide" "rpmfusion-free" "rpmfusion-nonfree" ; do for PROVIDE in "mariadb" "mariadb-devel" "mariadb-connector-c" "mariadb-connector-c-devel" "pkgconfig(mariadb)" "pkgconfig(libmariadb)" "mysql" "mysql-devel" "community-mysql" "community-mysql-devel" "pkgconfig(mysql)" ; do echo -e "\n\nREPO: $REPO ; PROVIDE: $PROVIDE"; dnf -q --repo="$REPO" --repo="$REPO"-source repoquery --whatrequires "$PROVIDE" --alldeps | tee "./RESULTS/$REPO-$PROVIDE" done done It show that other than the last few occurrences described in the bugzilla above, all is ready. '''Rename package 'community-mysql' to 'mysql'''' and '''Stop providing 'mysql' symbols by package 'mariadb': ''' When MariaDB was introduced to Fedora, it seemed like it eventually replaces MySQL and therefore the packages were designed so that 'mysql' names were provided by MariaDB, and MySQL was renamed to community-mysql. The both projects were drop-in replacements at that time, so that design helped the smooth transition. Since then, a lot changed. First, those two databases developed different sets of features. The last "major major" version of MariaDB marked as a drop-in replacement is MariaDB 5.5. From MariaDB 10.0, it is usually still easy to upgrade from one to another. https://mariadb.com/kb/en/mariadb-vs-mysql-compatibility/ Since MySQL 8.0 and circa MariaDB 10.5, the differences grew significantly and so it does not make sense anymore to provide 'mysql' names (= 'mysql' RPM `Provides:` ) by MariaDB package(s). Moreover, the package name 'community-mysql' is Fedora specific. Oracle upstream uses 'mysql-community', Debian, RHEL and CentOS Stream use 'mysql'. From my experience I can say users usually search for 'mysql' name. To keep the naming convention closer to Fedora-related clones, the proposal is to rename the MySQL packages from 'community-mysql' to 'mysql'. This change will save me, the maintainer, noticeable amount of time and energy when cherry-picking commits and patches from Fedora to CentOS Stream and RHEL. And the more energy I save downstream, the more I can put into Fedora and upstream. This change was being prepared to be proposed back for Fedora 37, but was left unfinished due to capacity reasons: https://fedoraproject.org/wiki/Changes/MariaDB_MySQL_Renaming '''Drop cross-installation functionality:''' In Fedora, it is currently possible, on the packaging level, to cross install server of one DB with client of another. Specifically, you can install MariaDB server with MySQL client or MySQL server with MariaDB client. I introduced this functionality to try it out, in hope of delivering a handy enhancement for the users. Sadly, the drawbacks out-weights the positives. This behavior became a generator of elusive bugs I was never able to resolve. E.g.: https://bugzilla.redhat.com/show_bug.cgi?id=2026933 Moreover, as the MariaDB and MySQL projects diverged significantly (see the above point), it stopped to make sense around the time the MariaDB stopped to be drop-in replacement for MySQL. AFAIK, Fedora is the only distribution I know of, which ever allowed such installation combination. I though it would be interesting, I tried it, I found out, now I want to end the experiment. This change should not affect any other packages from the stack ('mariadb-connector-c', the ODBC, Python or JAVA connectors, ...) as they implement the API on their own and don't rely on the client application. '''Switch to the versioned layout of MariaDB and MySQL packages:''' I used the modularity heavily, as a package maintainer. It allowed me to ship alternative versions of MariaDB that I or the users wanted. During chnages of the default major version of MariaDB in Fedora, they served as a handy way for anyone to use them way before the change is implemented, or after it when the haven't adapted their systems to the new default version. I've regularly selected the versions I maintain in CentOS Stream and RHEL - those which received the most care - and supported them fro the longest time as Fedora modules. With Modularity retired https://fedoraproject.org/wiki/Changes/RetireModularity this ability has been lost. When researching alternative approaches, only one seem to cover the same goals without too much of additional fuss. That is the "Multiple packages with the same base name": https://docs.fedoraproject.org/en-US/packaging-guidelines/Naming/#multiple I want to change the packaging structure so the result will look as follows: * The unversioned name ('mariadb') will become a meta-package ** It will point to the one versioned variant which I choose to be the default one for the given Fedora release ** It will provide all of the unversioned names for the versioned variant that is default for the given Fedora release, to minimize the changes visible to the users * All other versions will have their own versioned package (e.g "mariadb10.5" "mariadb10.11") and will conflict with each other This will allow for: * users to keep using the unversioned names they are used to * maintainer to change the default version for a given Fedora release on a single, centralized place * users to enjoy all of the features of the modularity I offered them, in a simpler way * maintainer to add new versions quickly, without any need of changing the default version (other than adding new conflicts) Note: I specifically don't want the packages to be parallel installable. I only want them to be parallel available. That's why I didn't choose "Alternatives" as the solution: https://docs.fedoraproject.org/en-US/packaging-guidelines/Alternatives/ '''Introduce MariaDB 10.11 and MySQL 8.1:''' I would like to introduce MariaDB 10.11 and MySQL 8.1 into the Fedora. Based on the previous point, it should be achieved without any user disturbance, as I won't change what is the default version in the distribution, I will just add an alternative for the users to choose from. Proof of concept (needs additional work) for MariaDB 10.11 here: https://koji.fedoraproject.org/koji/taskinfo?taskID=102927697 https://src.fedoraproject.org/fork/mschorm/rpms/mariadb/commits/10.11-LTS '''Change the default MariaDB version in Fedora from 10.5 to 10.11:''' To keep being the leading edge distribution, I propose to update the default version of MariaDB in Fedora to the latest upstream LTS version, the MariaDB 10.11. I have received a feedback on several occasions from various users that they would love to see this version in Fedora. Since I made a working proof of concept, I am confident that I will introduce the new version in time. The MariaDB 10.6 LTS release has been skipped (was only available as a module) in Fedora, due to strategic reasons. At the time of early days of MariaDB 10.5 and 10.6, the MariaDB upstream changed their strategy of delivering the new releases. They switched to predictable quarterly releases, with new "major major" release every quarter. With this change, all releases are short-term releases with a support for only a single year. Some of the releases are made LTS, mostly based on the community feedback, discussion and demand. This all was new and unclear at the beginning, so we selected 10.5 LTS to be the main version maintained through Fedora, CentOS Stream and RHEL, so users from all three operating systems would contribute via reports to the health of the release and all those user's MariaDB would receive the same care. When the 10.6 was announced to become LTS, we didn't rush for it, both due to capacity reasons and to keep the bigger user-base advantage. We didn't know how long it would take for the next LTS to be announced. Now, when the situation is clear, and the 10.11 was announced as the LTS, I believe it is appropriate to upgrade to this version in Fedora (and work is being started downstream to again keep the bigger user-base advantage). If the previous two changes are accepted and implemented, this change will be only a matter of updating the 'mariadb' meta-package to point to the 'mariadb10.11' as the Fedora default, instead of the 'mariadb10.5' == Feedback == (05/10/2023) Added clarification regarding other packages from the stack, namely connectors for C, ODBC and JAVA (05/10/2023) Added clarification that by "mysql names" I mean "'mysql' RPM `Provides:`" (05/10/2023) Added explanation for the specific order of the changes in this document == Benefit to Fedora == '''Drop builds for i686 architecture''' * Maintainer regaining time and energy, when I stop dealing with i686 specific bugs * Fedora infrastructure regaining the system resources needed to be put into hours long compilations, storage, etc. * Also 'BuildRequires:' regarding MariaDB and MySQL will get fixed across the packages '''Rename package 'community-mysql' to 'mysql' ''' and '''Stop providing 'mysql' symbols by package 'mariadb'''' * Aligning with other distributions * Recognizing the differences between MariaDB and MySQL that grew over time * Maintainer regaining some time and energy from downstream work '''Drop cross-installation functionality''' * Elusive bugs caused by this will vanish, maintainer doesn't need to deal with them * Recognizing the differences between MariaDB and MySQL that grew over time '''Switch to the versioned layout of MariaDB and MySQL packages''' * Allows to regain all of the functionality lost after Modularity retirement ** Helps maintainer to avoid the painful decision which single version of the DB to pack into Fedora ** Helps with upgrades of the default version - users may choose to upgrade earlier, or later '''Introduce MariaDB 10.11 and MySQL 8.1''' * New stuff that users wait for, yay \o/ ! '''Change the default MariaDB version in Fedora from 10.5 to 10.11''' * Update to a latest upstream LTS version, awaited by the users == Scope == * Proposal owners: ** request several repositories to be created ** may need to go through a package review process several times (unfortunately) * Other developers: ** ~6 remaining package(r)s to accept the PRs regarding usage of correct "BuildRequires:" or fix their packages in other way accordingly, in order to accomplish the i686 drop. * Release engineering: [https://pagure.io/releng/issues #Releng issue number] ** shouldn't be needed ? * Policies and guidelines: N/A (not needed for this Change) ** '''Should''' the Guidelines be in conflict with a reasonable approach I propose regarding the versioned packages, lowering their MUST to SHOULD in certain cases or getting FeSCo exception. * Trademark approval: N/A (not needed for this Change) * Alignment with Community Initiatives: == Upgrade/compatibility impact == '''Drop builds for i686 architecture''' * Users relying on i686 build of the database servers don't have any upgrade path '''Rename package 'community-mysql' to 'mysql' ''' and '''Stop providing 'mysql' symbols by package 'mariadb'''' * The package 'mariadb' will no longer be installed preferably instead of package 'mysql' when doing "dnf install mysql" (and other sub-packages) '''Drop cross-installation functionality''' * Users relying on this atypical setup don't have any clear upgrade path ** Such cases should only happen on development setups, which is solvable with containers or similar semi-isolation '''Switch to the versioned layout of MariaDB and MySQL packages''' * Nothing I can think of '''Introduce MariaDB 10.11 and MySQL 8.1''' * Nothing I can think of '''Change the default MariaDB version in Fedora from 10.5 to 10.11''' * https://mariadb.com/kb/en/changes-improvements-in-mariadb-106/ ** TokuDB has been removed *** As a result, the sources can be stopped to be modified downstream to strip problematically licensed code, and pure upstream tarball can be used instead ** The utf8 character set (and related collations) is now by default an alias for utf8mb3 rather than the other way around. It can be set to imply utf8mb4 by changing the value of the old_mode system variable * https://mariadb.com/kb/en/changes-improvements-in-mariadb-107/ ** New UUID data type * https://mariadb.com/kb/en/changes-improvements-in-mariadb-108/ * https://mariadb.com/kb/en/changes-improvements-in-mariadb-109/ * https://mariadb.com/kb/en/changes-improvements-in-mariadb-1010/ ** --ssl option set as default for mariadb CLI * https://mariadb.com/kb/en/changes-improvements-in-mariadb-1011/ == How To Test == '''Drop builds for i686 architecture''' * i686 builds doesn't exist anymore '''Rename package 'community-mysql' to 'mysql' ''' and '''Stop providing 'mysql' symbols by package 'mariadb'''' * The package 'mariadb' will no longer be installed preferably instead of package 'mysql' when doing "dnf install mysql" (and other sub-packages) * Using names 'mariadb' and 'community-mysql' leads to the same results as before '''Drop cross-installation functionality''' * Cross-installation not allowed anymore '''Switch to the versioned layout of MariaDB and MySQL packages''' * Test updates, upgrades, re-installs, etc. ** Users should get the same functional results as before with the same names as before, while having differently named packages present on the system '''Introduce MariaDB 10.11 and MySQL 8.1''' * Test upgrades / downgrades between versions '''Change the default MariaDB version in Fedora from 10.5 to 10.11''' * Test upgrades / downgrades between versions == User Experience == '''Drop builds for i686 architecture''' * i686 functionality lost (server only) '''Rename package 'community-mysql' to 'mysql' ''' and '''Stop providing 'mysql' symbols by package 'mariadb'''' * The package 'mariadb' will no longer be installed preferably instead of package 'mysql' when doing "dnf install mysql" (and other sub-packages) ** Otherwise should not be noticeable to the user '''Drop cross-installation functionality''' * Cross-installation functionality lost '''Switch to the versioned layout of MariaDB and MySQL packages''' * Users should get the same functional results as before with the same names as before, while having differently named packages present on the system * Functionality lost by Modularity retirement recovered '''Introduce MariaDB 10.11 and MySQL 8.1''' * New stuff that users wait for, yay \o/ ! '''Change the default MariaDB version in Fedora from 10.5 to 10.11''' * Latest upstream LTS version, awaited by the users ** https://mariadb.com/kb/en/upgrading-from-mariadb-10-5-to-mariadb-10-6/ ** https://mariadb.com/kb/en/upgrading-from-mariadb-10-6-to-mariadb-10-11/ == Dependencies == Only the i686 removal should be blocked by other packages. All other packages should keep working as they do. == Contingency Plan == '''Drop builds for i686 architecture''' * I'll try in F41 again '''Rename package 'community-mysql' to 'mysql' ''' and '''Stop providing 'mysql' symbols by package 'mariadb'''' * Revert '''Drop cross-installation functionality''' * Revert '''Switch to the versioned layout of MariaDB and MySQL packages''' * Revert '''Introduce MariaDB 10.11 and MySQL 8.1''' * I'll try in F41 again '''Change the default MariaDB version in Fedora from 10.5 to 10.11''' * Revert to the MariaDB 10.5 as the system default, but keep MariaDB 10.11 parallel available in the repository. Fix issues and try in F41 again. == Documentation == Let this document and the linked resources be the documentation. == Release Notes == Will be added later in the process, according to the changes applied -- Aoife Moloney Product Owner Community Platform Engineering Team Red Hat EMEA Communications House Cork Road Waterford

1 0

F40 Change Propsal: BogofilterSqlite (Self- Contained)
by Aoife Moloney 16 Oct '23

16 Oct '23

Wiki Link: https://fedoraproject.org/wiki/Changes/BogofilterSqlite Discussion.fpo link: https://discussion.fedoraproject.org/t/f40-change-proposal-switch-bogofilte… This document represents a proposed Change. As part of the Changes process, proposals are publicly announced in order to receive community feedback. This proposal will only be implemented if approved by the Fedora Engineering Steering Committee. == Summary == Switch bogofilter to use SQLite as its database engine, rather than Berkeley DB (libdb). == Owner == * Name: [[User:mikep| W. Michael Petullo]] * Email: mike(a)flyn.org == Detailed Description == Switch bogofilter to use SQLite as its database engine, rather than Berkeley DB (libdb). Another change (https://fedoraproject.org/wiki/Changes/Libdb_deprecated) marked libdb as deprecated, and that change lists bogofilter as a dependency. Thus this change fixes another application to avoid the deprecated libdb. == Benefit to Fedora == Fedora will have one less dependency on the deprecated libdb package. Additionally, other distributions have already migrated to SQLite, and this will allow sharing word lists with those distributions. For example, perhaps a workstation running Fedora generates wordlist.db before installing it on another computer running Alpine and acting as a server. == Scope == * Proposal owners: Merge pull request https://src.fedoraproject.org/rpms/bogofilter/pull-request/2. This makes the database backend conditional, with SQLite being the default. Support for libdb can be conditionally compiled to create a migration tool capable of migrating existing libdb databases to SQLite. * Other developers: N/A (not needed for this Change) * Release engineering: N/A (not needed for this Change) * Policies and guidelines: N/A (not needed for this Change) * Trademark approval: N/A (not needed for this Change) * Alignment with Community Initiatives: Support [https://fedoraproject.org/wiki/Changes/Libdb_deprecated Mark libdb as deprecated] == Upgrade/compatibility impact == Bogofilter can support only one database backend at a time, and thus a new SQLite bogofilter package will be unable to process libdb data. Thus the new package provides a migration script. == How To Test == This test generates a word list and migrates it to work with the new SQLite backend. Install original bogofilter and add at least one word to its database, for example with: echo abc | bogofilter --bogofilter-dir=/tmp/bftest/ --register-spam Bogofilter will create the directory `/tmp/bftest/`, and it will contain a `wordslist.db` file. To verify the word had been added run: bogoutil -d /tmp/bftest/wordlist.db Install the updated bogofilter and migrate the existing libdb database with: bogomigrate-berkeley /tmp/bftest/wordlist.db This tool will print whether the migration succeeded. Verify the "abc" word is present in the newly created SQLite database with: bogoutil -d /tmp/bftest/wordlist.db == User Experience == == Dependencies == N/A (not needed for this Change) == Contingency Plan == * Contingency mechanism: N/A (not a System Wide Change) * Contingency deadline: N/A (not a System Wide Change) * Blocks release? No == Documentation == N/A (not a System Wide Change) == Release Notes == The bogofilter package switched its database engine from Berkeley DB (libdb) to SQLite because Fedora deprecated libdb. Users can migrate their word lists manually with `bogomigrate-berkeley ~/.bogofiler/wordlist.db`. -- Aoife Moloney Product Owner Community Platform Engineering Team Red Hat EMEA Communications House Cork Road Waterford

1 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

devel-announce October 2023