ECOOP 4th referee's comments

A. Does the paper contain sufficient information to be evaluated?

        (x) Yes
        () No

B. Overall recommendation:

        () Strong Accept
        () Accept
        () Weak Accept
        () Weak Reject
        (x) Reject
        () Strong Reject

C. Confidence in your rating:

        () Highly confident
        (x) Confident
        () Uncertain, although familiar
        () Uncertain, because of unfamilarity

D. Does the paper fit the theme on Object Orientation?

        () Well
        (x) OK
        () Not at all

E. Tick the main field of the paper:

        (x) Programming Languages
        () Specification Languages, Semantics,  Formal methods
        () Analysis, Design, Methodology
        () Libraries, Frameworks, Patterns
        () User Interfaces
        () Tools
        (x) Concurrency
        () Application, Simulation
        () Metrics
        () Teaching
        () Paradigm Combination
        () Other: Distribution

F. Is the paper a research contribution or an experience report?

        (x) Research paper
        () Experience report

G. How do you rate the scientific quality/originality of the paper?

       () Good
       () OK
       (x) Low
       () Not acceptable

H. How do you rate the quality of presentation, English etc.?

       () Good
       (x) OK
       () Low
       () Not acceptable

J. Your comments to the author(s): Paper No:

The paper is very easy to read but it is written in a very superficial way and does not convince the reader that the approach proposed (typed transactions) actually enhances incremental development.

To be convincing, the author should have detailed his approach and presented a detailed example, together with a comparison with alternatives (e.g a CORBA approach).

The motivation for incremental development is very well stated.

The distinction between the two ways along which types are abstract is very confusing (page 5). Should explain it more clearly.

Blocking is the underlying DEFAULT model in CORBA. You might use also non-blocking calls.

The explanation why RPC restricts incremental development is not convincing at all !!!

Table of contents