2009/6/26
Since SCREAM OpenESL will be announced this month, I decide to have a new blog for the open sources developed at SCREAM Lab. For such information, please go to :
http://screamlabopensource.blogspot.com/
I would expect that new SLIM will be released soon.
--------------------------------------------------------
For some reasons, I decide that we will make our ESL tools open. They include our past efforts on Eclipse based SystemC IDE with GUI, Matlab/Simulink Co-simulation, FPGA Co-simulation, and all other heterogeneous mixed-level tools.
Before the 0.1 version is announced, there are some features to be added.
1. TLM 2.0 compatible interfaces.
This includes Buses and interfaces to heterogeneous tools. This will be taken care of by Buffett, ruru, and Garystone.
2. Matlab/Simulink and FPGA related examples and codegen and change of communication models from socket to shared memory.
This will be taken care of by Buffett and DNA.
3. Profiling tools and its visualization
This will be taken care of by ruru and Garystone.
4. Integration of GtkWave and Icarus (or Modelsim)
This will be taken care of by long, Buffett and Garystone.
5. User menu, technical menu, powerpoint and book.
This will be taken care of by Garystone, ruru, long and Alvin.
Let me know if I miss anything and there is something wrong.
Let me toss some options that should be done after the first version is released
1. Debugger under Eclipse
2. 8051 and DSP16 utilities
Last night, I felt so relieved and free that I made everything clear. There is no need to make 700K NTD while losing our freedom to make the best ESL tools in the world. AND, it is open and free.
11 則留言:
Finally, It is INDEED OpenESL now!
I think TLM2 support is the most important part now.
There are some minor utilities we can try 2 do :
1. Projectfile gen & manager
2. Grammar-Based Codegen
3. C++ -> OpenESL XML converter
今天騎車回家的時候，突然想到一件事，竟然前輩們(ex. al, keiko...)都覺得design pattern很好用，又容易維護，那我們是不是應該在以後的projects都儘量把這種設計思維、方法加進去，讓程式可以共好維護，且更有可讀性？
另一方面，也可以讓大家學學新把戲，不要老是寫Code都像我一樣混亂，當然如果有心，要再搞搞UML分析是更好的啦...
小小建議
Glad to hear another OPEN project to be released from SCREAM. In my opinion, open source project stands for freedom of development. It derives from the very first interest on the project itself. Something to be noted henceforth:
1. To make it really open, adverts such as websites or project hosting, should be really taken care of. Try google Code, sourceforge on which SLIM was hosted is getting somewhat worse.
2. Use cases, examples, and FAQ on OpenESL should be really clearly shown, so as to draw people to try it and afterwards improve it.
It was my fault that I did not make SLIM website on sourceforge "attractive" enough for others to give a try. Hope OpenESL will be SCREAM's first successful open source project. Hopefully worldwide known :P
Starry,
We will have SLIM re-engineered. This time, we have a bigger thinking.
In the past, SLIM is weak in not having many components to use. This time, we will try to integrate OpenMax and try to use SLIM as the IL and AL. Also, IDE on Eclipse will be up first. You are welcome to join us if you want. Let us know what we need to know and do for a great open source. Please go to the new SLIM thread to have your opinions posted. Also, OpenMax and Cell are two threads you can watch.
I think you did a great job in SLIM. Your SLIM is a starting point for OpenESL and this new SLIM.
I plan to give an undergraduate course for ESL. This may create quite a few examples for OpenESL. As for SLIM, I may ask graduate students to make components like monthly homework.
As you said, doing open sources is to own freedom of development and mind. Also, released us from those who gives money. This means that SCREAM will become poorer, but let me imagine. Happier.
BTW, let's also try to make SLIM well known to the world too.
DNA:
can you explain "Grammar-Based Codegen "?
Ohoh, it seems that English is more pouplar in this discussion. So, let me show you my pool English...
Using design patterns and UML diagrams is a good way to explore the world of software. But, it takes a long time to develop taste of your own to software design. I said this is a taste because software design has element of art :) However, if Prof. Alvin decides to train students for software design/development, I'll be very glad to share my poor experience and stories with SCREAMers >///< The sharing topics may involve design pattern, software architecture, UML (only rough) and, of course, how to apply/implement it in C++ XD
PS. No charge for the sharing. (It means Alvin doesn't need to pay me NT$100000 per month XD)
Oh, don't forget the words: "anti-pattern"
I strongly agree with Buffett. It seems that OpenESL is a suitable target to apply Software Engineering. An example is helpful for me to learn SE. It will be an instructive experience if engineers of TREND MICRO can help us. Let's facilitate SE, not just a slogan.
In my opinion, code generators should generate code according to the theories in compiler design. In contrast, a codegen spamming string-printing functions would be hard to debug or modify(If the codegen grows large enough). But maybe it's still not the time yet. Maybe we can try this way when generating TLM 2.0 socket codes.
Recently, I will ask my member in Computer and Network Center, NCKU, to teach ikki how to use a SW called EA for SE. It includes UML, too. I think I can just ask him to make the presentation in the lab. All are welcome.
Actually, I have been asking all members to follow but they seems not so appreciate. The reasons I do not want to explain here.
However, it is really important to make good projects. As researchers, we really do not have to do this. But, to make good SW, this is important. I think AL, Royce, and Keiko should agree with this.
Let's try it.
D.N.A.
I think these "FUTURE" features are what distiguish OpenESL from Coware.
Thx for your input.
張貼留言