Needs an abstract. What is an example of a real-time system with Best-Effort. Is it still real-time? 3.1.4 Talks as if no system has application level processing, and then goes on to list two. Resource reservation CPU-scheduling was used but not defined. ====================================================================== Good things: ------------ * Quite a bit of information on RTOSes. Good definitions of some of their key features (hard/soft real time, dependence on hardware). * Interesting expositions on the integration of GPOSes and RTOSes, the need for it, etc. * Paper was generally well-written and flowed logically. As the reader I got the sense that there was a variety of options in RTOSes, and there are interesting approaches to solving their unique problems. I also saw how RTOSes have become a current issue with hardware improvements. Therefore I felt that the paper achieved its goals. Improvements: ------------- Paper's Structure: * The authors needed to better prepare the reader for what he is going to read in parts of the paper with more clear/thorough introductory sentences. Many sections I didn't feel adequately prepared for what lay immediately ahead. * Parallel structure of comparisons could have lent itself to tables. Also please explain in an introductory section why some OSes were not discussed in that section (more preparation). * Didn't define the differences between processes/tasks/threads. * Many ideas would come up suddenly, or trail off suddenly, without explanation. * Could have provided some (minimal) numbers on the requirements of RTOSes for service deadlines, as compared to average timeslicing in a GPOS (i.e. a process is guaranteed to run in XX ns after an interrupt occurs, versus XX ms for a GPOS, on average). Just to make things more tangible. (How catastrophic is a page fault??) * Don't introduce new information/ideas in the conclusion. Many of the ideas presented in the conclusion looked good -- they could have been more clearly stated as your thesis, and focused on in the comparisons. Grammatical, etc: * several sentences were awkward; took me several times to understand some of them. * found several typos and word omissions * (too) many acronyms used and defined, not all were explained ====================================================================== I was not convinced of the paper's importance. There are a few too many acronyms to keep straight. The paragraphs did not flow well together. A large number of ideas were introduced but never explained. New ideas were introduced in the conclusion. ====================================================================== Your comments on the paper. This is public comments that the authors of the papers will see. Provide feedback to improve their paper, etc. Most aspects of your discussion were very clear, in particular the sections on memory management and the integration of RTOS' and GPOS'. My only negative comment was that you didn't focus much on connections or trends at a level larger than the seven design issues you discussed. At the end you briefly touch on the trend of increasing generality in RTOS'; it would have been nice to hear a little more about this. ====================================================================== Well done! Kudos to Anderson, et al.! This paper is well organized, well thought-out, and well written. One minor criticism: try to reduce the number of acronyms! ====================================================================== The paper distinguishes the differences between the requirements of the general purpose operating systems and the real time operating systems. It surveys several RTOSs and explores them with respect to the design issues that should be taken into account. For each of the design issues, these RTOS examples are observed in detail and the way they meet the main requirement, executing and completing tasks in a deterministic time frame, is presented. However, the content is generally informative, i.e. more criticism could be added. Also, future directions could be added in the end as an another section. ====================================================================== IMPORT: In my view, I think their subject area is of importance to the OS community, especially, in the light of emerging applications like multimedia which would require real-timeness from operating systems. However, I also believe that a lot of work has been done in the area of RTOS's since their inception. NOVELTY: This paper mostly seems to me to be a summary of the RTOSs and their mechanisms in the areas of scheduling, IPC, memory mgmt etc. The paper expresses no thoughts about any unexplored design areas that could be explored further, or any variance in the implementation of surveyed RTOSs which they feel could have made it better. Also, there isn't any mention of any possible further research areas in the RTOSs. QUALITY: I think that some of the observations made with regards to disk reads, memory management and scheduling were good. However, I think there is still a lot room available for giving critique for the surveyed RTOSs. OVERALL: I think that overall, the paper has been written well, in terms of presenting the salient features of the surveyed RTOSs. However, I believe it could have added more strength to the paper, had the authors brought forth more opinions/criticisms on the design considerations for the surveyed RTOSs. In addition to the neat presentation of issues, I felt that there is a scope for improvement in the order of presenting the ideas. e.g in the section on "Integration of RTOS and GPOS", it would have been great, if this section was opened describing the need for it first (which actually happens in the second paragraph) ====================================================================== Import - Real-time OS are an interesting topic and definitely need to be observed, for trends and good and bad features. Novelty - There did not appear to be any observations from the papers. It appeared that the information presented was more just what was found about each sub topic on the Operating systems. There were no conclusions drawn from the information researched. The title even said 'Comparison' while there was not really much comparison, that was left to the reader. The facts about the similar topic (hardware support, predictable memory management, ...) where presented very clearly (for the most part), but the comparison was lacking. Quality - There did not appear to be many observations, the information presented appeared to be correct. Again, there was not conclusions or comparisons of the information that was presented. Overall - The paper was well partitioned. Each section that presented the way that the OSs did dealt with the design issues (Hardware support, or Predictable memory management, ...) could have had a comparison, or at least an opinion paragraph, about what was good and bad about the different approaches, which one is the best, or under what circumstances would one be better than another. That would have show some classifications or attempts to analyze the information found in the research of the different Operating systems. All of the analysis was left for the reader to do. ======================================================================