mw 发帖数: 525 | |
m*****e 发帖数: 4193 | 2 He is talking about threading in traditional languages (which is achieved by
libraries instead of native language support), and arguing for better
language support for writing concurrent programs.
【在 mw 的大作中提到】 : http://www.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-1.pdf : 居然反潮流的说threading不好,这个倒是蛮独特的观点,大家怎么看?
|
g*****g 发帖数: 34805 | 3 Bullshit, I agree concurrent coding on threads can be difficult.
But until one builds a system that can scale at least as well and
also simpler in the coding, it's just nonsense.
Language is to make simple things simple and difficult things possible.
Make first part simpler is not enough to cut it if you can't
make the latter part possible.
【在 mw 的大作中提到】 : http://www.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-1.pdf : 居然反潮流的说threading不好,这个倒是蛮独特的观点,大家怎么看?
|
h*******e 发帖数: 225 | 4 这观点没什么独特的,看看多少年前Berkeley的SEDA吧。其实都是argue来argue去,师
兄写篇event driven好,发个sosp毕业了,师弟再推翻了,也发篇sosp毕业。
看看里面用数学来argue multi-threading不好的那一段,挺搞笑的。
【在 mw 的大作中提到】 : http://www.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-1.pdf : 居然反潮流的说threading不好,这个倒是蛮独特的观点,大家怎么看?
|