Re: [boost] Status of FSM viewer
Hi Andreas, thanks for the prompt answers. See my comments bellow. On Thu, 09 Sep 2010, Andreas Huber wrote:
we are looking for a solution to visualize FSMs in our robotic applications. Before starting the development of our own tool I'd like to ask whether there exists something able to visualize Boost.Statecharts FSMs.
No, as far as I know.
I was only able to find the following GSOC project: [snip links] but in my quick tests it produced only empty figures. I believe that after some debugging I could make it work.
Personally I wouldn't bother. IIRC, the project depends on GCC_XML, the latest release of which is still 0.6, from 2004. There is activity in source control but I'm not sure it has anything to do with the tool.
My question is whether the development of this tool continues or whether there are some other tools. In case of negative answers, do you have any suggestions for the development of the new tool? Currently, we consider the development of a GCC plugin which will generate dot (graphviz) source.
I don't know anything about GCC plugins, but I guess that is certainly an option. Personally, I'd also check out clang, which presumably can provide you with a much more C++-like representation of the parsed source code.
I had also thought about using clang but I was not able to compile even simple FSMs with it. I produced a lot of strange errors. I guess that clang's C++ support is not as mature as of GCC. Do you have more positive experience with clang++?
P.S. If you make your tool public, I'd love to be informed when you have a first version to check out. Thanks!
A student of mine will work on this in his bachelor thesis so the tool will definitely be public. If we have something that works, I'll let you know. Cheers, -Michal
I had also thought about using clang but I was not able to compile even simple FSMs with it. I produced a lot of strange errors. I guess that clang's C++ support is not as mature as of GCC. Do you have more positive experience with clang++?
When did you test clang? Quite recently, Doug Gregor announced that clang now compiles all of Boost. Also, the current test results look quite promising to me: <http://www.boost.org/development/tests/trunk/developer/statechart.html>
P.S. If you make your tool public, I'd love to be informed when you have a first version to check out. Thanks!
A student of mine will work on this in his bachelor thesis so the tool will definitely be public. If we have something that works, I'll let you know.
Excellent. Thanks, -- Andreas Huber When replying by private email, please remove the words spam and trap from the address shown in the header.
On Sep 9, 2010, at 1:56 AM, Michal Sojka wrote:
On Sep 9, 2010, at 4:28 AM, Andreas Huber wrote:
I had also thought about using clang but I was not able to compile even simple FSMs with it. I produced a lot of strange errors. I guess that clang's C++ support is not as mature as of GCC. Do you have more positive experience with clang++?
When did you test clang? Quite recently, Doug Gregor announced that clang now compiles all of Boost. Also, the current test results look quite promising to me:
<http://www.boost.org/development/tests/trunk/developer/statechart.html>
When Doug talks about clang compiling all of boost, he doesn't mean the latest released version (2.7, I think). He means taking the "top of tree" from the clang/llvm subversion repo. -- Marshall P.S. A new release of llvm (2.8) is scheduled for the end of September.
On Thu, Sep 9, 2010 at 9:39 PM, Marshall Clow <mclow.lists@gmail.com> wrote:
On Sep 9, 2010, at 1:56 AM, Michal Sojka wrote:
On Sep 9, 2010, at 4:28 AM, Andreas Huber wrote:
I had also thought about using clang but I was not able to compile even simple FSMs with it. I produced a lot of strange errors. I guess that clang's C++ support is not as mature as of GCC. Do you have more positive experience with clang++?
When did you test clang? Quite recently, Doug Gregor announced that clang now compiles all of Boost. Also, the current test results look quite promising to me:
<http://www.boost.org/development/tests/trunk/developer/statechart.html>
When Doug talks about clang compiling all of boost, he doesn't mean the latest released version (2.7, I think). He means taking the "top of tree" from the clang/llvm subversion repo.
I can confirm this, with template heavy code to boot. Yes, clang can build all of Boost, especially code that uses MPL and Spirit quite heavily, but only if you build clang from the trunk.
-- Marshall
P.S. A new release of llvm (2.8) is scheduled for the end of September.
I can hardly wait for clang+llvm 2.8! :) -- Dean Michael Berris deanberris.com
participants (4)
-
Andreas Huber -
Dean Michael Berris -
Marshall Clow -
Michal Sojka