I've made this change. Welcome, new mergers!
The following is repo specific info. It should live somewhere else, perhaps
in
https://github.com/wildfly/wildfly/wiki, but today it's here. (Some of
it may make sense for other repos, but I'm not talking about those here, so
don't assume I am.)
RULES FOR MERGERS to wildfly/wildfly
1) Only one person should be merging PRs at a time. We control this via
messages in Zulip in the
https://wildfly.zulipchat.com/#narrow/channel/274477-Pull-requests/topic/...
topic.
Before merging, check the thread and if no one else is merging post:
merging on <branch> :lock_with_key:
When done, post
:unlocked:
2) Avoid merging to a branch the week of its planned release until the
release is tagged (usually a Wednesday). Let the coordinator for the
release drive merging. If something really needs merging, discuss first in
the release's topic
https://wildfly.zulipchat.com/#narrow/channel/424923-releases channel in
Zulip.
3) PRs should have satisfactory reviews from people competent in the areas
it touches and it should have recent "good" CI results before being merged.
There are a lot of words in that sentence that involve exercising good
judgement. We're not highly prescriptive and rely a lot on good judgement.
If you're not sure, ask, or just don't merge.
Some details about use of judgement:
a) "Satisfactory" reviews. As much as possible this means a formal GitHub
approval. If someone has made a simple suggestion and you can see the
author followed it but you can't get that reviewer to formally approve in a
timely manner, then it's 'satisfactory'. "Timely" is a matter of
judgement.
b) "Competent in the areas it touches" is not defined. There is no formal
rule. The wildfly-bot will request reviews from people based on its config,
but that's just a suggestion. If the author requests that certain people
review, respect that, unless they tell you it's ok not to.
c) "Recent" CI generally means no more than a week old. Maybe less
depending on where we are in the release cycle and what other things have
been getting merged since the CI jobs ran.
d) "Good" results means all tests executions ran (so no failures that
stopped prevented executions running) and with no new failures or
potentially relevant intermittent failures.
Intermittent failures are a complex topic; too much to get into here.
Bottom line: use your judgement and if you're at all unsure, ask about it
in the main zulip channel
https://wildfly.zulipchat.com/#narrow/channel/174184-wildfly-developers
4) PR titles should mention the JIRA(s) so they end up in the merge commit
message.
a) Exception: dependabot PRs that don't affect production code or tooling
likely to be of interest to users (e.g. Arquillian, Galleon) don't need
JIRAs.
5) For dependabot PRs that impact production code, at some point I'll
research what modules are affected and post a comment @ pinging the
relevant people asking for input by date XXX, and otherwise it will be
merged. If you see one of these and its past date XXX and someone hasn't
responded, it's ok to merge.
If I haven't added one of these comments, ask before adding one yourself.
There are some components where we really want an explicit approval and
silence isn't enough.
6) Don't use "/retest" just to get a new run of one or two failed CI jobs.
Don't waste energy. If you don't have TeamCity perms to click on the link
for a failed job and hit the "Run" button to make it run again, talk to Ken
Wills.
7) Don't merge feature PRs unless you understand the feature development
process well and are very clear that the requirements have been met. When
in doubt, ask.
8) No merging of feature PRs between the tag of a Beta and the opening up
of the branch for the next feature release following the tag of the Final.
9) No merging of PRs with the ##.x label on them if ## is later than the
release under development.
Best regards,
Brian
On Fri, Sep 19, 2025 at 1:36 AM Jean-Frederic Mesnil <jmesnil(a)ibm.com>
wrote:
On 18 Sep 2025, at 22:05, Brian Stansberry via wildfly-dev <
wildfly-dev(a)lists.jboss.org> wrote:
I'm proposing we significantly expand the set of people who have write
permissions to the wildfly/wildfly GitHub repo. The following folks are all
experienced WildFly developers who act as a component lead for components
in the WFLY JIRA project, but they don't have write permissions to the main
server code repository:
Tomasz Adamski
Jean Francois Denise
Emmanuel Hugonnet
Rado Husar
Tom Jenkinson
Jason Lee
Scott Marlow
Eduardo Martins
Brad Maxwell
Matěj Novotny
Richard Opalka
Harald Pehl
Flavia Rainone
Marco Sappe Griot
Big thumb up for this list.
Jeff
--
Jeff Mesnil
Software Engineer
Red Hat JBoss EAP
http://jmesnil.net/
IBM
Unless otherwise stated above:
Compagnie IBM France
Siège Social : 17, avenue de l'Europe, 92275 Bois-Colombes Cedex
RCS Nanterre 552 118 465
Forme Sociale : S.A.S.
Capital Social : 664 614 175,50 €
SIRET : 552 118 465 03644 - Code NAF 6203Z
--
Brian Stansberry
Architect, JBoss EAP
WildFly Project Lead
He/Him/His