Delivery-Date: Thu Nov 20 08:23:35 2008
Message-Id: <200811201620.mAKGKr1e020236@bourbon.usc.edu>
To: cs551@merlot.usc.edu
Subject: Re: CS551_Final2_index files
Date: Thu, 20 Nov 2008 08:20:53 -0800
From: Bill Cheng <william@bourbon.usc.edu>

Someone wrote:

  >    [BC: Added 11/19/2008]
  >    The "name_index" and "sha1_index" must be sorted.
  > 
  > I've also updated the grading guidelines to add the sorting
  > requirement to name_index and sha1_index.
  > 
  > For sort, I sort name_index by key which is filename, and it looks like
  > 
  > 1
  > blondie1.mp3
  > 4
  > blondie1.mp3
  > 2
  > chess.jpg
  > 3
  > usctommy.gif
  > 5
  > usctommy.gif
  > 
  > and same manner for sha1_index
  > 3
  > 11CDDF2A0378D6A892CF43198847CE379E85D149
  > 5
  > 11CDDF2A0378D6A892CF43198847CE379E85D149
  > 2
  > F7917FE4976D2C24F225BC4B6C2334E554B91C28
  > 1
  > FFD3B197E1C0F0C27E7BDC219F553A3E6B139DFB
  > 4
  > FFD3B197E1C0F0C27E7BDC219F553A3E6B139DFB
  > 
  > I guess this should be ok, right?

You need to have a "conflict resolution chain" in your
index.  Therefore, I would have:

    blondie1.mp3,1,4
    chess.jpg,2
    usctommy.gif,3,5

and:

    11CDDF2A0378D6A892CF43198847CE379E85D149,3,5
    F7917FE4976D2C24F225BC4B6C2334E554B91C28,2
    FFD3B197E1C0F0C27E7BDC219F553A3E6B139DFB,1,4
--
Bill Cheng // bill.cheng@usc.edu <URL:http://merlot.usc.edu/william/usc/>


  > On Wed, Nov 19, 2008 at 10:27 PM, Bill Cheng <william@bourbon.usc.edu>wrote:
  > 
  > > Someone wrote:
  > >
  > >  > Can we simply use structures to store the names,SHA1 values and
  > > bitvectors
  > >  > of all the files present on a node.
  > >  > Although it is given in the Spec to use linear list and binary search
  > > tree.
  > >  > Can't we create object of a structure whenever a file gets stored on a
  > > node
  > >  > and when a search is to be performed, we can search through all the
  > > objects
  > >  > to find a match. This won't be efficient but atleast correct, Right ?
  > >
  > > The spec explicitly asks you to implement these structures.
  > > To make it even more clear, I've just updated the spec to
  > > include the following:
  > >
  > >    [BC: Added 11/19/2008]
  > >    The "name_index" and "sha1_index" must be sorted.
  > >
  > > I've also updated the grading guidelines to add the sorting
  > > requirement to name_index and sha1_index.
  > >
  > >  > It is given in the spec that 'The file names must be "kwrd_index", "
  > >  > name_index",  "sha1_index". These files are disk images of the
  > > corresponding
  > >  > memory index structures.'
  > >  > Is there any fixed format of how these files should look like ? Or can
  > > we
  > >  > simply use any format
  > >
  > > Correct.
  > >
  > >  > Like for kwrd_index file, it contains
  > >  > bitvector1
  > >  > bitvector2
  > >  > .
  > >  > ..
  > >  > ...
  > >  > bitvectorN
  > >
  > > Well, they need to be "correct".  In your example, you
  > > have include at least the corresponding file number.  So,
  > > it can look like the following:
  > >
  > >    bitvector1 fn1
  > >    bitvector2 fn2
  > >    .
  > >    ..
  > >    ...
  > >    bitvectorN fnN
  > >
  > > where fn# refers to a file number in the HomeDir/files
  > > directory.
  > >
  > > Similarly, for name_index, you can use an ASCII format
  > > for it.  For example:
  > >
  > >    filename1,fn11,fn12
  > >    filename2,fn21,fn22,fn23,fn24
  > >    .
  > >    ..
  > >    ...
  > >    filenameM,fnM1
  > >
  > > There can be multiple file numbers per filename since
  > > multiple files may have the same original FileName.
  > >
  > > You can also do the same thing for sha1_index.  So, they
  > > are actually quite easy to implement, especially because
  > > they are not required to be trees!
  > > --
  > > Bill Cheng // bill.cheng@usc.edu <URL:http://merlot.usc.edu/william/usc/>