On 10/3/2020 1:42 PM, Ishan Chattopadhyaya wrote:
As you might be aware, the reference_impl branch has a lot of
improvements that we want to see in Solr master. However, it is
currently a large deviation from master and hence the stability and
reliability (though improved in certain aspects) remains to be tested in
real production environments before we gain confidence in bringing those
changes to master.
I propose that we do a one off preview release from that branch, say
Solr 10 alpha (early access) or any other name that someone suggests, so
that users could try it out and report regressions or improvements etc.
How to handle this seems to come down to the answers to a couple of
questions:
* Is this new code stable enough to work in the wild?
* Do we want to release 9.0 before we merge to master, or after?
Can someone who was involved when 4.0-ALPHA and 4.0-BETA were released
tell me whether those releases actually did what was intended and made
4.0.0 better than it would have been without the special releases? If
they did, then a special release for this new implementation before
merging to master would probably also be helpful. If there was no real
benefit gained, then maybe we're better off just going ahead with the merge.
If the general feeling is that this new release is looking very solid,
then I think we should merge to master soon, probably just before
branch_9x is created. If we think it needs more work, then maybe we
should hold on merging until *after* branch_9x is created, so the new
implementation will release with 10.0 and there will be more time to
make it bulletproof.
My current impression, which I will admit is made with almost zero
actual information about the code or its stability, is that we should
plan on stabilizing the new implementation for the 9.0 release. There's
going to be some pain no matter how we handle this, so diving in now
seems like a good idea.
A tangent: How robust is our testing? I know they take a long time to
run, which Mark's new implementation aims to improve, but do we think
there's enough being tested? There has been some discussion in the past
about benchmarks for Solr. Benchmarks would be very useful, but I'm
more interested in making sure that our tests will catch problems before
they get out into the wild. The two goals don't have to be mutually
exclusive, though.
Thanks,
Shawn
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
