Well, thats an interesting statement with lots of implications.
As a software developer, your job when diagnosing problems is to be an investigator. To figure out why something is crashing or why something is slow requires collecting of evidence, analyzing the evidence, and ultimately trying to apply that back to the bug and symptoms. Sometimes, unfortunately, problems are hard to investigate, and even some pretty smart developers can come up with what I would call voodoo explanations. It used to be that people sometimes ran into “compiler bugs” or “optimizer bugs”. But these days, when a developer claims its one of these, its usually a pathetic, last-ditch effort to explain why his code is not working. The compiler bug is just pretty darned unlikely. In fact, I can’t remember the last time I saw a bug that was actually the compiler’s fault. I’m sure they exist, they are just realy rare. The reason they are unlikely is because there is so much code exercising the compiler that there is just a huge amount of testing on it. If there were a bug, lots of software, not just yours, would be broken.
Gmail’s claim that my browser cache being full also sounds like voodoo to me. I searched around to see if there were any other products claiming performance woes due to the cache, and I can’t find any. I also can’t find any support articles from Microsoft. Now, I could believe a turned off cache could cause problems. But logically, it doesn’t sound right to me that somehow the cache being full is the cause of Google’s troubles. The cache, when full, should rotate out the old content, and new content gets cached. Its a pretty simple algorithm, and if it were broken, wouldn’t most other web applications be suffering as well?
So, gmail, whats the nitty gritty? Do you have empirical evidence? Prove to us this isn’t voodoo engineering!