2012-01-31 22:38, David Holmes skrev: > I don't understand the comment on the new method though.
Since jre_release_version() has a similar use case as vm_release(), ie it is called from crash handlers and error reporting handlers, I copied the comment for vm_release() to jre_release_version(). But perhaps it is a bit too much information since it is not that likely that the code will be refactored to have stringStreams instead of char *. > I'm assuming that the actual problem is the ccache compares the > defines used for a compilation and so if the version string contains a > timestamp and gets updated on each (incremental?) build then ccache > thinks it has to recompile. Yes, that is correct. As can be seen in most build server logs, the ccache is rarely used. It is only used when a build number is explicitly set, since the build number prevents a date and time to be inserted in the JRE_RELEASE_VERSION. //Fredrik
