Some sort of filter based filesystem

Svein Halvor Halvorsen svein-freebsd-questions at
Wed Jul 20 22:25:28 GMT 2005

* Lowell Gilbert [2005-07-20 09:19 -0400]
>  You're right that the outputs wouldn't *have* to be pre-generated, but 
>  doing it on the fly would make the project both more difficult to 
>  implement and (I think) less convenient to use.

This, I do not get. Why is it, that this would be less convenient? In 
fact, I see no other use for this than to save disk space when you want 
data available in several (information perserved) formats?

>  It would be pretty slow if you really did generate it on the fly.  

This is correct, and this is why this would not fit for all cases. 
However, alot of the time, one could want to sacrifice speed for space 
and the convenience of having just one master dataset to maintain.

>  Also, a lot of other things would get very tricky: for example, you 
>  wouldn't have the file size available when listing a directory of 
>  "virtual" files.

This could be a real problem, I agree.

Svein Halvor

More information about the freebsd-questions mailing list