BDB Engine removal
« previous entry | next entry »
Aug. 24th, 2006 | 12:42 pm
Elvis has left the building, and yes we have removed the BDB engine from MySQL.
Last night I saw a blog post about the removal before I went to bed. Link to the post is here, along with my response:
http://www.webyog.com/forums//index.p
Removing the BDB engine, aka Sleepycat's Berkeley database, from the list of internal engines has been on my list to do for a while. If you follow the way I went about removing it, it should be easy to transform it into a plugin now.
Why as a plugin?
Right now we don't spend energy on keeping the BDB engine up to date. Which makes us look bad, and it makes Oracle look bad. What is worse is that the BDB engine used a forked version of the Sleepycat code. The BDB engine should not be using a forked version. The handler code for BDB, aka the shim, needs to be modified to use a standard version of the Berkeley code.
Now that the engine is no longer in the main tree no one has to work through MySQL to patch BDB. If BDB was well maintained the door might even be open for us to consider shipping it like we do other third party engines.
If you follow my commits from the last couple of weeks you will notice that I have been doing a lot of cleanup work. MyISAM, HEAP, and almost all of the engines have now been sealed in the storage directory as plugins. The goal is to make everything in the server work through interfaces.
Keeping the MySQL maintained engines behind the storage engine interface makes us eat our own dogfood :)
Last night I saw a blog post about the removal before I went to bed. Link to the post is here, along with my response:
http://www.webyog.com/forums//index.p
Removing the BDB engine, aka Sleepycat's Berkeley database, from the list of internal engines has been on my list to do for a while. If you follow the way I went about removing it, it should be easy to transform it into a plugin now.
Why as a plugin?
Right now we don't spend energy on keeping the BDB engine up to date. Which makes us look bad, and it makes Oracle look bad. What is worse is that the BDB engine used a forked version of the Sleepycat code. The BDB engine should not be using a forked version. The handler code for BDB, aka the shim, needs to be modified to use a standard version of the Berkeley code.
Now that the engine is no longer in the main tree no one has to work through MySQL to patch BDB. If BDB was well maintained the door might even be open for us to consider shipping it like we do other third party engines.
If you follow my commits from the last couple of weeks you will notice that I have been doing a lot of cleanup work. MyISAM, HEAP, and almost all of the engines have now been sealed in the storage directory as plugins. The goal is to make everything in the server work through interfaces.
Keeping the MySQL maintained engines behind the storage engine interface makes us eat our own dogfood :)
(no subject)
from:![[info]](http://p-stat.livejournal.com/img/userinfo.gif)
awfief
date: Aug. 25th, 2006 12:00 am (UTC)
Link
How exactly do you want it to be communicated?
Last I checked, you were supposed to read:
http://dev.mysql.com/doc/refman/5.1/e
before upgrading to a new version of 5.1. If you check out the 5.1.12 release:
http://dev.mysql.com/doc/refman/5.1/e
It says it right there, in the very first bullet point. Given that it's not even released yet, I don't know what you're complaining about. If you are using the latest nightlies then certainly you should not be surprised to find stuff broken as cruft is cleaned out; as well, you should read the commit logs if you're doing the nightlies.
But again, what do you have in mind, spam to all the lists and forums and e-mail addresses MySQL has, whenever a change is committed?
I do not think you owe anyone any reasoning or justification or even "don't worry, it can be put back!" The complaint was not "I can't believe they are taking BDB out," the complaint was "THEY'RE NOT TELLING US."
Except you are telling folks, that's why you have those web pages set up. Very very very few software products have as great release notes as MySQL does. They're very readable, and great to have.
Reply | Thread
(no subject)
from:![[info]](http://p-stat.livejournal.com/img/userinfo.gif)
rawy
date: Mar. 18th, 2007 02:50 pm (UTC)
Link
what about INNODB?
Reply | Parent | Thread
PgSQL gloats
from:![[info]](http://p-stat.livejournal.com/img/userinfo.gif)
qu1j0t3
date: Aug. 27th, 2006 10:15 pm (UTC)
Link
Reply | Thread
Re: PgSQL gloats
from:![[info]](http://p-stat.livejournal.com/img/userinfo.gif)
krow
date: Aug. 27th, 2006 11:40 pm (UTC)
Link
The author also points to the fact that BDB is not in the 5.1 manual, but fails to mention that it is still in manuals previous to that version.
I find the article pretty sad. There are a lot of good people working on Postgres and the wankers don't shed the best of lights on the entire project.
Reply | Parent | Thread
Re: PgSQL gloats
from:![[info]](http://p-stat.livejournal.com/img/userinfo.gif)
qu1j0t3
date: Aug. 28th, 2006 01:42 am (UTC)
Link
Reply | Parent | Thread
(no subject)
from:![[info]](http://p-stat.livejournal.com/img/userinfo.gif)
jaobedoza
date: Feb. 17th, 2008 06:29 pm (UTC)
Link
Reply | Thread