Christoph_Borchert

- friends
1 link karma
2 comment karma
send messagemember for
what's this?

TROPHY CASE

papers

subscribe8 readers

Welcome to the public review site of papers appearing in MobiSys 2012. Here you can find the public review written by a PC member for each paper and the authors' rebuttal to it. You can comment on any of them or add your own opinion. Finally you can vote to "like" or "dislike" a paper by clicking the up or down arrows next to its title.

CiAO/IP: A Highly Configurable Aspect-Oriented IP Stack by ChristophBorchertin papers

[–]Christoph_Borchert 0 points1 point ago

sorry, this has been archived and can no longer be voted on

This question was about the byte-order conversion via macros, as it is used in uIP, lwIP, Linux and BSD. Yes, this design practice is really bad, since a programmer has to aware of that all the time. If you access packet headers, you MUST NOT forget to surround your code by those macros (htons, htonl, noths, ntohl). This is of course for portability, but via AOP, the same portability is gained by a single aspect that fully encapsulates the byte-order conversion. AOP really simplifies portability in this case.

CiAO/IP: A Highly Configurable Aspect-Oriented IP Stack by ChristophBorchertin papers

[–]Christoph_Borchert 0 points1 point ago

sorry, this has been archived and can no longer be voted on

The answer to this question - how to be aware of changing base code that is affected by aspects (known as the fragile pointcut problem) - is actually to rely on IDE support. For Ecplise, there is an AspectC++ plugin that helps the developer to understand the interaction of aspects and ordinary C++ code. Obviously, it is NOT the file ID ;-)