Hi!
On Oct 10, 2008, at 9:31 PM, MARK CALLAGHAN wrote:
What happened?
I am disappointed with how 5.1's replication turned out. Both from a
quality perspective and from a code perspective, I am just
unimpressed. What exists in 5.1 is a hack on top of the work Sasha did
eight years ago. There was no thought to restructuring the code and
the way it links into THD annoys me. Look at how they didn't bother to
remove rpl_failsafe.cc dead code and further take a look at how they
left a bunch of dead 5.1 beta code in the execution tree (aka some
abandoned work that went wrong around the protocol).
From a quality stand point the fact that they reverted to using
statement as the default, back from mixed, tells me the mood on the
quality of it. It means they do not trust it to even be used a little.
Part of what we are trying to do is make something which is
maintainable going forward, I don't find that what was done in 5.1
makes that happen, I think it was a set back.
So we can roll back to 5.0's replication, write a new binlog, or try
to refactor what they have done. There are 96 active bugs between 5.0 -
> 5.1. A total of 131 bugs currently logged.
Do I really want to spend my time on it? Not really.
PBXT is the new Innodb. The main reason for doing this is that the
code we have has been tricky to get right. If we change it to use
There is a long blog entry that I will publish soon about this. I'm in
favor of what Innodb has done in the past, but I think their
development has crawled to a stop. Paul is more active which makes me
more inclined toward his work.
I'll write up sometime about this next week.
Cheers,
-Brian
internal XA then it is more likely to eventually make it into other
branches and the official release. It would also work for other
strorage engines and I am most interested in PBXT and Maria.
Where in the land of drizzle branches is the change to clean up the
net send/receive code so it uses blocking calls with timeouts
rather
than setsockopt(non blocking), try call, setsockopt(blocking),
request
a timeout alarm from another thread, do blocking call, ...? If you
have it, I want to copy it.
We do, but you would need to look at our patch history. It would be
simpler
to just have me show you were to remove the code from then try to
take a
patch. Will you be at the opensql camp?
No, I won't. I looked at net_serv.cc in the main Drizzle branch and
didn't not see a change there.
Cheers,
-Brian
Cheers,
-Brian
--
_______________________________________________________
Brian "Krow" Aker, brian at tangent.org
Seattle, Washington
http://krow.net/ <-- Me
http://tangent.org/ <-- Software
_______________________________________________________
You can't grep a dead tree.
--
Mark Callaghan
[EMAIL PROTECTED]
--
_______________________________________________________
Brian "Krow" Aker, brian at tangent.org
Seattle, Washington
http://krow.net/ <-- Me
http://tangent.org/ <-- Software
_______________________________________________________
You can't grep a dead tree.
--
Mark Callaghan
[EMAIL PROTECTED]
--
_______________________________________________________
Brian "Krow" Aker, brian at tangent.org
Seattle, Washington
http://krow.net/ <-- Me
http://tangent.org/ <-- Software
_______________________________________________________
You can't grep a dead tree.
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
