Thursday, November 11, 2004

The DOM of DOOM

When I first started working at Microsoft, my job was to implement the XML DOM for IE5. This was back while W3C was still defining DOM Level 1 and XML was a newborn babe. One of the guys I worked with (Rich Rollman) was on the W3C's DOM working group, and he and I would debate API issues over email at least once a day. The one thing that I remember most clearly from those days was that Outlook would always try and correct "DOM" to "DOOM". Little did I realize how right it was.

Fast forward 6 years. Now I manage the team working on DOM in the next release of the .Net Frameworks (code-name Whidbey). A few months back Sun released a parser/DOM performance benchmark (commonly called XML Mark). We have been investigating this benchmark and our numbers for Whidbey just were not as good as we wanted. The developer who owns our DOM code came into my office this morning to tell me that he found a bug in their tests that mean that meant that they never excersize their save code in Java, while the .Net version is correct. Fix this and our Whidbey numbers rock! Now that is how to make your Lead a happy boy!

Note: The bug is a very simple typo. I am making no claim that this was intentional on Sun's part and we are in the process of informing them of our findings. Competition is a good thing, and if this just spurs Sun to improve their DOM code even more, that only ends up benefit both of our customers.

1 Comments:

Anonymous Anonymous said...

And, did Sun acknowledge the problem and fix it?

11:38 PM  

Post a Comment

<< Home