One of our campus Blackboard Learning System Vista Enterprise administrators reported to have reduced the number of Java cache related issues (failed sessions) by changing the Disk Space Allotment from the 1,000 MB default down to 100 MB. This is found in the Java Control Panel > General tab > Temporary Internet Files: Settings. I am curious if anyone else has found this to be the case?

The purpose of web browsers having a cache was to speed up use of a web site by not having to download content again. RAM is faster than disk is faster than Internet. (This especially was true in the mid 1990s.) Take a look at this web site. There is the image at the top plus various CSS, and JS files. It looks like there are a good 224 KB in CSS, JS, and their supporting images. Rather than download significant amount of content again, with the appropriate settings a browser will check whether the size changed (assume no changes) or it expired (really that it is stale). If neither are true, then it uses what it already has. This will make my web site load faster for the user. So caching is a very good thing.

Java Plug-in, the client downloading and rendering applets in a web browser, works similarly. It can keep a copy of the applet in a cache. Starting with Java 1.3 there are even parameters placed in the HTML for applet caching. It looks to me like the HTML Creator, really edit-on(R) Pro by RealObjects, JavaScript for instantiating the applet has settings which enable Java to keep it in its cache.

The default cache size of 1,000 MB sounded excessive at first. Do people really reach the point where the whole cached is used? Looking at mine, I have 4 items in Applications from running them on my desktop plus around 2,200 items in Resources. All this takes up only 155 MB. Most of them are tiny files. The largest ones in Resources are from the various Vista  clusters I administrate. Therefore setting this to 100 MB as recommended probably means these getting downloaded more often and waiting on 1MB+ files to download. Glad we have a fast Internet connection at work. Sucks to be the students on DSL who follow this advice and use lots of Java-based applets.

If the Java Plug-in cache was buggy, then I could foresee problems with display of applets. It should download the applet but does not, it should not download the applet but does, the wrong applet is used, a corrupted applet is used. Instead, this seems to be claiming to solve an issue were the web browser lost the session cookie. It seems very unlikely to me that a Java Plug-in could cause a web browser to lose a session cookie much less changing the cache size fix it.