Re: Review of Thomas's >2GB ext2fs proposal

[Neal H. Walfield]

Thomas,

>  My proposal is better because it does these tasks in user space.

Your other comments basically made sense, however, in the context of
this statement, I have become extrememly confused.  So, before I react
to your other comments, I would like you to elaborate what you mean by
this statement.  Specifically, what about my proposal does not happen
in user space?  What I suggest (and what Ogi has basically
implemented) are changes to libpager (in the form of adding
pager_notify_eviction) and modifying ext2fs to only map active
metadata.  Nothing about this involves changes to the kernel; nothing
I propose involves using undocumented features; everything I suggest
functions within the Mach 3 framework.

Thanks,
Neal