Minor corrections

Gerald,

I've made corrections as follows.  I'm sending you these
such that we can agree on that which is correct; I can
then send you the corrected manuscript and that will
hopefully be 100% correct and we can go to print ASAP.

I will try to meet the 5th April deadline for submission
(such that I graduate in the summer), but that might be
a little steep.  Nevertheless,...


With regards,


K
================
p.11, para 2.	Ethernet has a maximum frame size of 1518 
bytes (not including preamble and start frame delimiter).
-> "... unit in contrast to Ethernets variable length
    packets of maximum 1518 bytes."

p.11, para 4	" path based of a small local identifier." 
Should read " path based on a small local identifier."
* Ok

p.21, last para	[7 lines from end]  " details of the 
filing system are not prescribed by the paper."  Should 
this be " details of the filing system are not described 
by the paper." ??
-> changed to "... though the exact structure of the file
   system is not dictated by the paper."
* They don't specify the fs in detail, since
  it is mathematically modelled.  Provided that the underlying
  fs can meet the model, then the technique can be applied.

p.44, fig 3.4 and 3.5.	Need to label y-axis with units.
* Ok, (but does Peter agree with my units?)
-> Now "Mean Disk Reads /\n # times frame read" and
       "Standard Deviation /\n # times frame read"

p.67, para 2.	" giving a maximum of 65535 f-nodes 
total." Should read " giving a maximum of 65535 f-nodes in 
total."
* ok

p.71, para 5.	Need to correct the definition of the ATM 
cell header.  The cell header itself does not contain the 
cell length information.
-> changed to "... The network header contains the VPI/VCI
    address pair, a cell loss priority bit, a payload type
    field, flow control information and a checksum."

p.86, section 4.4.3.1	check formula: "(5 x 4) x 25 = 
500".  Need to check if the 4 required, as discussed in the 
viva.
 -> changed to "... is assumed, 5 x 25 = 125 index entries
     per second is required.  Essentially, this is  index
     block per second of data  one index per 4 seconds of data."
* You are indeed correct - 5 disk blocks (of 4K) at 25 fps.
  I give both 'of a second' and 'in a second', since the
  latter unit is used later as a comparison: "... 13.6
  seconds of video per index block ..."

p.107, section 5.2.1	" by the addition on an element " 
should read "  by the addition of an element  "
* ok

p.119, para 2	printing error in equation. Reads "0 . efft 
< 1".  Should probably be "0 <= efft < 1"
* Trying 'Times Roman' instead of 'normal text' symbol.
  Works in my previewer, but a printer?  I'll test it later.

p.130, section 5.4.5	" of this few users in within the 
achieved performance "  should read " of this few users 
is within the achieved performance "
* ok

P.148, section 6.2.1	Need correction to description of 
comport.  This has an 8 bit wide data path.
-> The original "... Each comport is a bi-directional
    half-duplex asynchronous serial line running at ..."
    is now changed to " ... Each comport is a
    bi-directional half-duplex asynchronous 8-bit wide
    parallel port running  at ...".

p.155, para 2	" whilst network protocols all use 
big-endian byte order."  Change the "all" to "normally".
* ok

-== End of Corrections ==-

Here are some things I found :

p37 1st para, "there is question of" should be "there is
a question of".

p32 second para, ".. forward direction, but regions.. "
should be phrased ".. forward direction, but sections..".
The paper talks about regions on the disk; the original
wording could be confusing.

p30 bottom of page.  p = 3 and n = 4 is actually p = 3 and
n = 5.

p7 line 5.  'life' to 'lifetime'.



 