February 2004 - Posts
Joel has a top notch post on the GC in Rotor, it would be great to see some of Joels notes hooking into the code so that as he talks us through the GC we can see which parts of the Rotor code base (files, functions and macros) relate.
Interesting post from Chris Sells on his Ref counting project for Rotor, the good news is that it works well (removes the need for IDispose), the bad news is that Rotor is compiling very slowly when the code has the updated JIT added to it. In this paper (an overview of Ref counting and the reasons for adding it to the Rotor code) Chris states the slow down is occuring during the C# code stage of the build cycle (post PAL and post C# compiler), so the suspect may be the C# compiler its self and the GC, Chris has not been able to profile the heap to see just whats going on but when thats done it should give a good idea of whats tripping up. Hopefully at some point I will find the time to dig in.
The code can be found here (just a suggestion but it would be good to see this code on the Rotor community web site).
One book I am buzzed about is
Build Your Own .NET Language and Compiler by Edward G. Nilges, sadly not due out till May (and that means June/July for me) however I did manage to find an
article by Edward that has some good coverage of his QBasic compiler (used as a way of creating an eval command) and all written using VB.Net.
Miguel has a heads up on a new functional language for .NET called
Nemerle (
Microsofts F# is another functional language), works with
Mono, Microsoft CLR (it *should* work on Rotor with out a problem) and
P.Net.
Various J2EE bridges for .NET out there but Mains
oft have recently announced Visual MainWin for J2EE that works well in Vs.NET and compiles the MSIL to Java Bytecode. Not sure how well it would work with webforms or winforms but the
fact its using Mono code has me buzzed.
One of my research area's of late has been grid computing, during this research I came across the
Alchemi grid computing framework for the CLR. An opensource framework developed by researchers at the University of Melbourne.
Jonathan de Halleux has a great article on Code Project that shows his newest project
GUnit. GUnit differs to NUnit and csUnit in that it supports a much
wider range of unit tests. Would be great to see its source be opensource and the project grow and grow.
John Lam has a great link to an early mailing list post on why Microsoft chose deterministic finalization for the CLR.
For some reason my agg software has not been picking up
Yun Jins great posts on the CLR and Rotor so his posts (sadly) moved off my Radar, however a manual check this morning shows some great, great posts. If your a CLR wonk I highly recommend you check it out. How spoiled are we,
Michal,
Joel and Yun make up for some great Rotor/CLR posts !!
More Posts
Next page »