« The Open Source Job Market is Good | Main | Light Up The Darfur Wall »
December 5, 2006
Observations from Watching MySQL Source Code Commits
A few week ago we switched kruckenberg.com over to a new server. With the move I lost all of my procmail filtering rules which were responsible for automatically moving incoming mail to various folders on the server (mail must be filtered on the server so my Blackberry and webmail will get the same filtering).
Anyhow, one of the things I'd been sticking into a folder and looking at occasionally (meaning once a year to purge) is the messages from MySQL source commits. Anyone can subscribe to the list on the MySQL mailing lists page and get these messages which essentially contain the notes and diffs from when something is checked into version control. Most of these commits are for the database application, but there is also a steady stream of documentation changes coming in. Since I haven't gotten around to redoing the server-based filtering the MySQL source changes are popping up in my inbox with every other important and urgent email.
Over the past few weeks (since November 16th to be exact) I've been watching these a little closer because they are interesting to an outsider. I thought I'd throw a few observations out there.
- the code never sleeps: Code commits happen around the clock, there has rarely been a time when an hour passes without a commit. Based on when my inbox gets the most messages I'd say there are is a heavy concentration of development that happens in Europe because when I wake up at 6 or 7am my inbox is already overflowing. By early afternoon it has died down significantly.
- many hands...: I count 68 unique committers in this chunk of commits (1184 total commits).
- Monty *does* work on the code: I've heard numerous folks say that Monty (one of the MySQL founders) is very involved writing MySQL code. It is true. In fact, Monty is in the top 10 most frequent committers over this period of time.
- 5.1 is king: The highest percentage of commits were to the MySQL 5.1 code, 5.0 being not far behind. There is also a fair amount of work happening in version 4.1. Version 5.2 had a few changed and even 4.0 showed up once.
- who are these folks?: I know a good chunk of MySQL employees but most folks who've committed code in the past two weeks are folks I haven't met before. It is nice to see a familiar name from time to time like Jay, who had some commits for both the 5.0 and 5.1 tree recently.
- it's not all there: I have yet to see anything about Falcon, which I'm anxious to get my hands on - there must be some rules about what is and isn't sent to the list because I know there's development going on with Falcon but nothing of it in the source code commits
It has been interesting to watch, but I don't think it is going to last much longer as I think I'm ready to devote some time to setting up filters on the new machine.
Anyone else have comments on things they've noticed from watching these commits?
Posted by mike at December 5, 2006 10:34 PM
Comments
Falcon is in a separate tree. There are, well, many trees. Not everything makes it public, and definitely not everything is mirrored on bkbits either.
Posted by: colin at December 6, 2006 3:49 AM
Hi!
Yeah, you want to see a commit that I'm keeping my eyes on? Check out this one that Antony Curtis is working on:
http://lists.mysql.com/commits/16433
It's the first cut of making server variables dynamically tied to the plugin system, which means we finally will start to see all the #ifdef cruft being removed from the major /sql directory files. Check out the sheer number of #ifdef removals from mysqld.cc! Yeah! Yeah! Yeah!
Posted by: Jay Pipes at December 6, 2006 1:52 PM