> With the sfio copy, it will create holes if possible. it leaves it
> up to the file systems by doing an lseek to the offset of the next non-zero
> byte after encountering some number of 0 bytes.
Just to make Dave's point clear, sfio deals with a sequence of zero's
on output as follows:
1. It writes out as many zero's as possible until it achieves alignment
with some alignment value (page size or disk block size).
2. Then, it does lseek() forward as far as possible while retaining
alignment.
3. It writes out the rest of the data.
So, when a file is copied to a new file, some parts of the sequences of zeros
in the first file will become holes in whatever way the file system thinks of
holes.
>From an application's point of view, a file is always a contiguous sequence of
>bytes.
Holes are just an optimization done by the file system and Sfio allows the file
system
to choose how it wants to do that optimization.
Phong
_______________________________________________
ast-users mailing list
[email protected]
https://mailman.research.att.com/mailman/listinfo/ast-users
