Re: [ros-kernel]Registry status

[Casper H. <ch...@us...>]

tir, 2002-04-16 kl. 09:45 skrev Aliberti Emanuele:
> >3) It seems that when a new hive is created because it does not exist, 
> >CmiPopulateHive() is called to fill out the hive with 50 empty 
> >pages/blocks, growing the file to 200K.  Is that really necessary?  It 
> >would be nice to keep the files size down.  When I exported a small hive 
> >from 2k, it was only 8K.
> 
> For internal lab testing, we once booted a NT 4.0 box after completely removing the hives.
> The system goes up anyway and creates the full hive set, but the files are onyl 4 kbyte large. Shortly after, the system hangs (of course).
> A memory page just fits for the hive's header.

Yes, a newly created registry hive contain only the base block (the
first 4KB). CmiPopulateHive() is meant only for testing write support
(which is not there yet). Expanding a hive requires new functionality,
so it is easier if the hives do not need to be expanded when
implementing write support.