Referee 1 *******************************************

E: Referee Comments (For Author and Editor)
------------------------------


Good overall description of the Grid problem and definitions of Grid
portal. You might want to show some screenshots of GRB in operation.
Good description of services provided. In section 4, maybe another
diagram detailing the layering of the software required in the
implementation would be helpful- also a description of how the various
pieces fit together. For instance what role does Expect play with GRB?
Given that there are Java, Perl and Python approaches to Grid computing
emerging, what were the factors in determining C and CGI as a superior
approach?
I'd like to know more about the C libraries you mention in the
Conclusion section, possibly with a description in Section 6: project
status.
Overall I think this is a concise, well organized paper with valuable
informational content.


Referee 2 *******************************************

E. Referee Comments (for Author and Editor)

The subject and content of the paper are quite satisfactory.  The paper
does contain some grammatical errors which should be corrected prior
to publication.

F. Presentation Changes

A larger typeface and/or less interline-spacing (leading) would improve
the appearance.  Section headings should not appear as the last line
on a page.

Referee 3 *******************************************

E: Referee Comments (For Author and Editor)  ------------------------------

Overall the papers is good, but there are a few sections that are either incomplete
or inaccurate.  Section 2.5 says that GridFTP supports third-part transfers, which 
is true, and then goes on to say that the Globus toolkit doesn't support third-part 
transfers. This is incorrect. The Globus GridFTP client libraries do support third-party
transfers.

The paragraph on security in section 3 needs to be corrected and expanded.  It is not
true that MyProxy uses one-time passphrases. It uses a user supplied passphrase, and not
a one-time passphrase which is a technical security term. I'd also like to see more then
one sentence about how cookies are used. What information is stored in them? How are
they protected from malicious use?

In section 4 the claim is made that using C in their portal provides better performance. Is 
there any data to back that claim up? With the overhead of CGI, Java Servlets are
frequently benchmarked faster in industry. If this claim is going to be make, I'd like to 
see some performance data.

The section on Project Status could use some information on users of GRB. What 
application groups are using the portal? How well does a generic portal work to meet
the needs of multiple application groups?

F: Presentation Changes

The overall English reads a bit rough currently.