On Fri, Jul 10, 2020 at 2:12 AM <bugzilla-daemon@spamassassin.apache.org>
wrote:
> https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7826
>
> --- Comment #34 from Henrik Krohns <apache@hege.li> ---
> (In reply to Kevin A. McGrail from comment #32)
> >
> > Please remember with votes being made on this thread, you should be
> waiting
> > 72 hours for people to weigh in.
>
> People have weighted in.
>
Which is why I have started working again. :-)
> Again, who is "we"? Why would _trunk_ have any effect in your "two
> systems"?
> Use a branch as suggested, so no other change in trunk will bother your
> testings etc. And people will actually have something to weight in, before
> things break!!
>
We is myself and my staff. The changes others made on this bug to trunk
had an effect because we already had fixes ready to commit but we had to
backtrack them.
A branch is a bad suggestion due to the number of testers we need and it
doesn't test ruleqa/masscheck.
From the way this bug started, I have little faith in the way you are
> going on
about it. Is there even a plan?? Where is it?
Of course there is a plan. See the top of the bz 7826 for the high-level
plan which is effectively this:
Anything referencing whitelist will be welcomelist (it was originally
allowlist but the PoC led to a good recommendation to use that).
Blacklist to Blocklist, Master to Parent, Slave to Child.
backwards compatibility will work for at least a year and until at least
4.1 is released.
all backwards compatibility will be documented as deprecated
The first commit was a proof of concept for just one function
(whitelist_to) to see what would break and how it worked in the live
system. It broke worse than expected because I forgot a part of the commit
but led to the good idea of using welcomelist.
I've just committed a second version of the proof of concept for
whitelist_to. If it looks good, next up is whitelist_from. From there,
rinse and repeat until done.
I recognize that people are using trunk and were upset by this so I will be
more careful on commits.
> What things are you going to
> rename?
Anything that references the terms above.
> What things are you going to break? "I'm going to rename WLBL to
> ALBL... uhh no it's not necessary to rename it..",
It was necessary to rename it originally because Whitelist was becoming
Allowlist so ALBL made sense. The idea of Welcomelist, however, is perfect
and removes that need.
> how about actually posting
> the complete plan and waiting for people to weight in on all the renames?
Let me know if the plan above is not sufficient. Happy to give you more
insight and discuss for consensus. In my defense, I write voluminous notes
and plan things out.
> Or do
> as suggested and go on implementing the things in [ed: a branch] so we
> have review and
> test the changes easily, instead of yet again more questions in mailing
> lists
> on why there are errors and why rule updates are not working etc.
>
Sorry but no, a branch will just hide issues. We need to find bugs.
However, I really think the rough part is over and we have a repeatable
roadmap. PoC 1 had a mistake in the commit. PoC 2 looks clean and will be
mimicked for the continuation of the work.
Regards,
KAM