[debian-mysql] Request for release team decision on MySQL and MariaDB [was: Re: Bug#793316: Bug#793316: transition: mysql-5.6]

Clint Byrum spamaps at debian.org
Mon Jan 11 18:59:22 UTC 2016


Excerpts from Pedretti Fabio's message of 2016-01-11 05:52:26 -0800:
> 2016-01-07 15:49 GMT+01:00 Norvald H. Ryeng <norvald.ryeng at oracle.com>:
> 
> > On Fri, 18 Dec 2015 22:31:05 +0100, Robie Basak <robie.basak at ubuntu.com>
> > wrote:
> >
> > 20:13:13 <pochu> 3- we have two forks of the same codebase
> >>
> >
> > That's not something upstream can do anything about. And if having
> > forks is a problem, why wasn't that issue raised when MariaDB was
> > accepted into Debian?
> >
> > Debian carries other forks, e.g.:
> >
> >  - GNU Emacs and XEmacs
> >  - djbdns and dbndns
> >  - FreeMind and Freeplane
> >  - Nagios and Icinga
> >
> > I don't see any reason why MySQL and MariaDB should be any different
> > from other software.
> >
> 
>  Note however that for other debian packaged softwares a choice was made,
> here are some I just remember:
> - openoffice / libreoffice
> - xfree86 / xorg
> - libav / ffmpeg
> 

Great point, that both situations exist.

> Actually mariadb and mysql can be used as alternative, but what should do
> in the future packages depending on them when they'll start to diverge?
> Specify just one of them as an explicit requirement? But then some packages
> depending on one cannot be installed along packages depending on the other
> because mariadb and mysql cannot installed at the same time?
> Or just use a set of common features compatible with both, possibly being
> unoptimal with both? And which one should users install?
> 

They've already diverged enough that they're only alternatives on
the network port, low level protocol, and 99.9% of the SQL language
alone. They cannot be used interchangeably on the same host without
custom migration work. Though MariaDB I believe still can be migrated
to directly from MySQL 5.6's data files, config values may be different,
and so on.

There are a myriad of reasons to choose either one from a user
standpoint, and, perhaps most importantly, there are actual developers
standing at the ready to do the work to support that choice.

> I think that in the interest of Debian a choice should be made, as it was
> recently made with the libav / ffmpeg alternatives. Also other distros
> already did a choice. Users needing the other can still install from a 3rd
> party repository.
> 

And all of this likely won't go away any time soon, Because there is such
a deep valley between the communities that sit on top of the two projects.
The fact that MySQL has a very unfriendly development model drove those
other distros to MariaDB, and it drives a lot of community involvement
that way. But on the other side, that unfriendly development model has a
_ton_ of money behind it, and keeps MySQL out in front on some areas
that are critical for some users. Enough users exist on both sides, that
the choice is -very- important.

There are no other large distributions that offer the choice to
_developers_ quite like Debian. Are these big projects? yes. Is it a
burden on security and release? Of course. But does that mean we can
just close our eyes to one side of the community or the other? I think no.



More information about the pkg-mysql-maint mailing list