Hi Otto,
I wanted to respond to your proposal for the server team to own this. Sorry 
this took a while, I have discussed this with various internal folks to ensure 
I’m not communicating anything inaccurate.

Due to my request in the MIR meeting on Matrix [0] to already review the
case despite the ownership question not yet being sorted out, we already
have that reply above. Thank you Pushkar for the swift processing. The
review left the ownership question open for us to continue to discuss
here.

BTW - it is really great that you brought this topic up as it forced us
to outline what would be needed to make it possible some day - which is
part of the MIR process design & intention.


# 1 Duplication

MariaDB and MySQL have diverged somewhat in features over time but the 
functionalities are still similar enough to potentially violate our duplication 
rule. 
Yet one could also say at a higher abstraction level Postgres is the same and 
that lives alongside it in main.
And at a fine detailed level the MySQL and MariaDB of 2025 are no longer as 
much “the same” as they used to be.
Still, for now, MariaDB + MySQL would in my reading still violate the 
duplication rule [Line 29-49 in [1]].

As they continue to diverge into the future this becomes less and less of an 
argument.
We can and should get rid of the other blockers over time, and by then I’m 
quite confident that the duplication aspect will not be the one that eventually 
blocks us.


# 2 Need to move it to Main

What is the need to have it in main?
As outlined above there are alternatives for relational databases in Main - 
MySQL is just one.
There are further Databases in the Archive that live a happy life in Universe 
with Community support.
After all you are yourself the best example of that going well, we highly 
appreciate your collaboration on this with our Team in Ubuntu and Debian alike.

A commitment to maintain something in the Main component is always a nice step 
up in regard to quality control, bug tracking and bug processing for our users.
But it also represents quite an investment. It is a bet on the future need of 
our users as we’d thereby further commit to it for at least the next 12 years.
To be able to do so just comes with the requirement that we should all probably 
have a bigger maintenance team behind both solutions.
Probably on both of our sides before supporting two such databases. Right now I 
sadly can’t back that sufficiently for what it needs and deserves.

As Mohamed outlined back when we discussed MariaDB as default [2], for now 
MariaDB still has a smaller user base compared to MySQL.
We want to allow users to choose - which they already can, but need to avoid 
negatively impacting the MySQL user base because of MariaDB.
Given the extra investment it represents for us I’d be afraid that it would 
take focus away from MySQL.

With the appearance of Ubuntu Pro, covering a longer time as well as
packages in universe, these lines got blurred further and made the
former, more common, need to put something into Main less required these
days.

All that combined makes me not yet fully say “oh yeah it totally needs
to be in Main” while at the same time not having the resources to do
that well.


# Co-operation (of people for sure, but even more so DBs)

To me the most significant blocker to moving MariaDB to Main is that
MariaDB and MySQL are not really co-installable or co-operatable. That
is IMHO worth resolving independently of moving MariaDB into Main some
day or not, for the benefit of the overall ecosystem.

As I stated above they are evolving away from each other, there is a
future when something may only operate well with one or the other. I’m
afraid of a future where one package picks one over the other as
incompatible dependency. A package doing so would indirectly become
mutually exclusive with a package choosing the other and thereby causing
a derivative conflict. I’d really like to avoid that future for the
benefit of our users. And therefore, until that day, the two need to
become cooperative.

Let us ignore the past and try to let them become two truly different, 
co-operatable solutions.
Some aspects to resolve:
The shared paths on disk
Port usage conflicts
library name conflicts
I know that right now I can only outline these issues and I’m aware that I do 
not yet have a solution to propose.
Some might be challenging, for example who is allowed to keep the default paths 
and ports?
There might be more to be identified as that continues, this is the set that 
came to my mind so far.

I consider becoming truly non-conflicting a blocker for both directions
by the way - potentially getting MariaDB into Ubuntu-Main, as well
potentially as having MySQL back in Debian again some day. And I think
that would be beneficial for everyone no matter what else happens.


Thank you in advance for continued cooperation until this fully works out and 
further beyond!
Christian

P.S. While talking with many folks Mohamed’s raised another idea maybe
worth exploring: Since your intention is to further promote MariaDB it
might be interesting to consider creating a MariaDB snap (as Mohamed’s
team does for MySQL based on the deb that is in the archive). That would
bypass most of the problems related to the co-existence of MySQL and
MariaDB. And as a bonus further remove dependencies on a specific
Debian/Ubuntu version.

[0]: 
https://matrix.to/#/!xYmIdubccwfYWURVHS:ubuntu.com/$Qn0dCjIuYnuBVQPNJ61rRr1hzS-GrHYwSceKMj0xO8Q?via=ubuntu.com&via=matrix.org&via=stusta.de
[1]: 
https://canonical-ubuntu-project.readthedocs-hosted.com/MIR/mir-reviewers-template/
[2]: https://lists.ubuntu.com/archives/ubuntu-devel/2024-October/043165.html

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2122095

Title:
  [MIR] mariadb

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/mariadb/+bug/2122095/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to