SystemC Blockdiagram

[giselher]

SystemC Blockdiagram (without wiring)
Wiring will come separately, because I still have my hard time figuring out how to get the right informations for displaying ports and signals.

• Contains only svg stuff that is needed for displaying the SystemC Blockdiagram.
• Contains the STL implementation I pulled from Kevin

I think it is for the better to send this pull request now, because you can test while the others are still trying to fix their STL implementation.

[giselher]

I forgot to tell that it takes some time for diagrams of bigger designs to show up.

[rainerf]

Some early feedback:

• I'm getting lots of QPixmap: It is not safe to use pixmaps outside the GUI thread errors on the command line as soon as I start a SystemC simulation, even before a simulation context is even found.
• I'm also getting some QSortFilterProxyModel: invalid inserted rows reported by source model errors (again before a simulation context is found). This might be unrelated, but I never noticed those with the current master.
• I got the following traceback (once gain before a simulation context is found):

Traceback (most recent call last):
  File "/home/rainer/svn/ricodebug/src/plugins/SysCDiagram/SysCDiagramPlugin.py", line 126, in run
    object_vec = self.ctx["m_child_objects"]
TypeError: 'NoneType' object has no attribute '__getitem__'

• Please don't use upper case in file names (SVGImage.py)
• svgview.py and htmlvariableview.py are nearly identical. This is unacceptable.
• The same goes for svgview.mako and htmlvariableview.mako.
• The simulation context plugin is still broken (probably due to the problems introduced by the pretty printer?).

Saving the best for last, though: The good news is that, for all designs I tried the plugin on, it did not crash and the output produced looked very good!

[rainerf]

Something's wrong here, line 108 also has a function update(self).

[giselher]

Could you please test this new set of commits again? I simplified everything I could.

[giselher]

Rebased ontop of vermaks newest branch.

[rainerf]

7f390fc crashes for me upon loading the module:

  ...
  File "/home/rainer/svn/ricodebug/src/helpers/pluginloader.py", line 66, in __togglePlugin
    if not self.pluginldr.loadPlugin(self):
  File "/home/rainer/svn/ricodebug/src/helpers/pluginloader.py", line 119, in loadPlugin
    pluginmodule = __import__("plugins.%s.%s" % (plugin.modulename, plugin.classname))
  File "/home/rainer/svn/ricodebug/src/plugins/SysCDiagram/SysCDiagramPlugin.py", line 25, in <module>
    from datagraph.svgview import SVGDataGraphVW
ImportError: No module named svgview

[giselher]

Oh, sorry missed a file to add.

[rainerf]

Just got this crash while sc_get_curr_simcontext() was not (yet) available in any scope:

Traceback (most recent call last):
  File "/home/rainer/svn/ricodebug/src/plugins/SysCDiagram/SysCDiagramPlugin.py", line 113, in update
    object_vec = self.ctx["m_child_objects"]
TypeError: 'NoneType' object has no attribute '__getitem__'

[rainerf]

Other than the traceback above, drawing the diagram still works nicely!

[rainerf]

It's reproducible on my machine: just debug some program that does not
use SystemC (or stop a SystemC simulation on main() where
sc_get_curr_simcontext() does not yet exist) with the plugin enabled.

Python version is also 2.7.3 here.

[rainerf]

Fixed in 9862372, thanks.

[rainerf]

Another thing I just noticed: self.block_diagram.write_raw("bd.gv") will leave a temporary file in the working directory. Also, it will most probably crash if the working directory is not writable. Could you maybe use something like http://docs.python.org/2/library/tempfile.html?

[giselher]

Oh, that file was just for debug purposes. This line is not needed any more.

[rainerf]

Ok. is this ready to merge then? Will the further fixes for the pretty printer be rebased on this?

[giselher]

This is ready for merging.

I have WriteAccess to every repo and rebased it ontop of rainerf/master.
See the following comment on daniels pull request.
#48 (comment)

[rainerf]

Merged into integration branch (3c16ab2), closing.