Re: [MinGW-dvlpr] MinGW Installer

[Hin-Tak L. <hin...@ya...>]

Keith Marshall wrote:
<snipped>
> I was thinking in terms of unpacking to a virgin local directory, then using 
> `du' to report the aggregate size of all allocated disk blocks, but if I do 
> that on `ext3' I'll get a smaller total than on Win32, firstly because `ext3' 
> has a smaller physical block size, and secondly because it allows block 
> fragmentation, so sharing free space in partially filled blocks among 
> multiple files.  (It also does defragmentation `on the fly', to maintain peak 
> performance at all times).
<snipped>

(Apology my first post ever is off-topic) I am quite sure ext3 does not do 
'defragmentation on the fly'. The answer is in an FAQ - I believe on the e2progs 
site -: the linux kernel reserve the next x blocks (x=16?)
whenever possible during a write to a ext2/ext3 partition, so they are
far less likely to get fragmented, but they do. In fact a tool called e2defrag
exists (linked from e2progs), and I have personally used it myself a few
years ago. There are also at least one commercial ext3 defrag programs I
know of as well.

The last time I run e2fsck on one of my ext3 partitions it was shown as being
10% fragmented - that partition is where I run a particular p2p program, so it 
has a lot of large and sparse files which are gradually "filled"
out-of-sequence. It says ext3 can be hugely fragmented under some usage, and 
there is no on-the-fly defragmentation. The only situation where one can say
on-the-fly defragmentation is when one runs ext3 on top of LVM or a software
RAID; but then the "defragmentation" comes from the LVM or RAID rather than
ext2/ext3.

Feel free to research on this and seek a 2nd opinion.

Hin-Tak



		
___________________________________________________________ 
Try the all-new Yahoo! Mail. "The New Version is radically easier to use" – The Wall Street Journal 
http://uk.docs.yahoo.com/nowyoucan.html