22 Mar
2005
22 Mar
'05
3:46 p.m.
Mike Nordell wrote:
If it seems I come down a bit hard on the proposal, I
can only blame
language barriers. My intention is only to illuminate some of the pitfalls I
think are inherent in a dynamic load system as the proposed.
I apologize again. The intended proposal was meant for Wine only. I
cc'ed ReactOS because there was a Question about ReactOS at the end. I
wanted to know if there was some codebase I could use. So the proposal
is not at all relevant to ReactOS Kernel registry! ;)
You did make good points about speed and Security. A few are relevant to
Wine, and I will take them into consideration. (For Wine).
By the way from tests I made on Windows some of the XML implementations
(DOM style, load everything to memory) are faster than Registry
functions. I suspect that is do to hash table size, and User-Kernel land
switching. but it is in the same time frame.
And... "dynamic load system"s is not a new subject. It is just the same
as a file system, with a mounting table and all that. Even the very root
uses a driver stack (Some what hardcoded) and lots of Unix systems know
how to switch roots mid-flight....If you ask me I'd put all that
Registry hell directly into the file system. Why do we need a duplicate
implementation. It is already dynamic, driver driven, secure, cached,
Journaled, optimized to death...But Microsoft made that mistake in the
past and we most follow.
Sorry about the miss-understanding
Free Life
Boaz