[savant-devel] Discussions from IRC channel #savant for week 48/2005

[Rudolf Matousek]

Discussions from IRC channel #savant (irc.oftc.net) for week 48/2005:
---------------------------------------------------------------------

2005-11-29 12:55:00  <Rudas> Hi Doug,
2005-11-29 12:55:26  <Rudas> have you seen my comments on mailing list?
2005-11-29 14:49:48  <dhickey> Hi Ruda, I have been working on getting the documentation fixed and started looking into the seg fault issues
2005-11-29 14:50:23  <dhickey> did you receive any errors on the test2.vhdl file when not using capture-comments?
2005-11-29 14:53:52  <Rudas> no
2005-11-29 14:54:13  <Rudas> actually when I run test2.vhdl without comments then it is ok.
2005-11-29 14:55:12  <Rudas> The only thing I did, I started the scram in DDD debuger and let it segfault.
2005-11-29 14:55:13  <dhickey> I received an error when using g++-3.4 so I've been setting up a g++-4.0 environment to test with
2005-11-29 14:55:22  <Rudas> I see.
2005-11-29 14:55:26  <Rudas> I'm using 3.4
2005-11-29 14:55:44  <Rudas> (since you have recommended it) :-)
2005-11-29 14:56:13  <Rudas> But I can switch to 4.0.
2005-11-29 14:57:04  <dhickey> Whatever works :)  I am not sure why I would be receiving an error on 3.4 if you're not
2005-11-29 14:57:49  <dhickey> are you working with the most current cvs?
2005-11-29 14:59:45  <Rudas> yes
2005-11-29 15:00:33  <Rudas> so you are getting errors from the test2.vhdl without the flag capture-comments?
2005-11-29 15:01:05  <Rudas> and if I remember right there were not many changes in CVS.
2005-11-29 15:01:22  <Rudas> the latest was related to the missing virtual modifier.
2005-11-29 15:01:27  <Rudas> (and now something in docu)
2005-11-29 15:03:12  <dhickey> I'll have to check my setup, I've been away from savant devel for awhile...
2005-11-29 15:03:27  <Rudas> ok.
2005-11-29 15:04:08  <Rudas> I can run the testing plugin once again and tell you where it crashes on my config...
2005-11-29 15:04:46  <dhickey> sure, that will help, I was seeing the seg fault with capture comments, but would like to make sure it's for the same reasons
2005-11-29 15:06:18  <Rudas> just a minute, DDD just started
2005-11-29 15:07:23  <Rudas> in std::string::length () from /usr/lib/libstdc++.so.6
2005-11-29 15:07:33  <Rudas> in IIRBase_String::get_text_length (this=0x0) at IIRBase_String.hh:55
2005-11-29 15:07:33  <Rudas>  in IIRBase_String.hh
2005-11-29 15:07:43  <Rudas> in IIRBase_TextLiteral::convert_tree
2005-11-29 15:07:54  <Rudas> backtrace in reverse order...
2005-11-29 15:08:27  <dhickey> I'll compare that to what I have and track it down
2005-11-29 15:08:29  <Rudas> once again:
2005-11-29 15:08:43  <Rudas> std::string::length() from libstdc++.so.6
2005-11-29 15:08:59  <Rudas> IIRBase_String::get_text_length() at IIRBase_String.hh
2005-11-29 15:09:24  <Rudas> IIRBase_TextLiteral::convert_Tree() at IIRBase_TextLiteral.cc
2005-11-29 15:09:43  <Rudas> IIRBase_Comment:convert_tree() at IIRBase_Comment.cc
2005-11-29 15:10:43  <Rudas> IIRBase_List::convert_tree() at IIRBase_List.cc
2005-11-29 15:10:55  <Rudas> IIRBase::convert_node() at IIRBase.cc
2005-11-29 15:11:13  <Rudas> IIRBase_DesignFile::convert_tree() at IIRBase_DesignFile.cc
2005-11-29 15:11:35  <Rudas> testing_plugin_interface::process_tree() at testing_plugin_interface.cc
2005-11-29 15:11:42  <Rudas> main() at main.cc
2005-11-29 15:11:50  <Rudas> (now it is complete)
2005-11-29 15:12:02  <Rudas> well, there are no line numbers...
2005-11-29 15:12:33  <Rudas> the get_text_length() is IIRBase_String.hh:55
2005-11-29 15:13:04  <Rudas> and the IIRBase_TextLiteral::convert_tree() at IIRBase_TextLiteral.cc:79
2005-11-29 15:13:12  <Rudas> I hope that will help.
2005-11-29 15:15:50  <dhickey> thanks, i'll check it out
2005-11-29 15:18:22  <Rudas> I just send a mail with snapshot of the backtrace window in the mailing list
2005-11-29 15:18:40  <Rudas> I put there my file test.vhdl - which should be a copy of test2.vhdl.
2005-11-29 15:18:49  <Rudas> It should, you know...
2005-11-29 15:37:30  <dhickey> my environment is back to normal now...I think I was experimenting with a couple things and never recompiled
2005-11-29 15:37:42  <Rudas> I see...
2005-11-29 15:37:47  <Rudas> okay.
2005-11-29 15:43:07  <dhickey> so far it looks like "text" is defined in IIRBase_Comment::convert_tree correctly, but IIRBase_TextLiteral::convert_tree calls get_text_length on a null text variable
2005-11-29 15:47:56  <dhickey> in IIRBase_Comment::convert_tree() it is calling TextLiteral::convert_tree which is then trying to set_text on the new_node which hasn't had the text set in the original IIRBase_Comment call
2005-11-29 16:32:37  <Rudas> ok...
2005-11-29 16:33:48  <Rudas> so you know how to fix it, do you?
2005-11-29 16:34:03  <Rudas> don't you?
2005-11-29 16:35:05  <dhickey> sure....
2005-11-29 16:35:40  <dhickey> I'm testing something right now...I'll post a patch if it works
2005-11-29 16:36:47  <dhickey> did you follow the call path I referred to?
2005-11-29 16:37:31  <Rudas> not now, I'm working on an article for a local magazine.
2005-11-29 16:37:50  <Rudas> If you think that I should have a look, then I'll do
2005-11-29 16:38:55  <dhickey> don't worry about it, hopefully this will take care of it
2005-11-29 16:39:02  <Rudas> thanks!
2005-11-29 16:39:29  <dhickey> no problem, I will post a patch to the list and update CVS when this is fixed
2005-11-29 16:40:06  <Rudas> cool, I'll update my code and check with my part plugin.
2005-11-29 16:40:29  <dhickey> I will also have the savant-devel-primer and a simple install manual available soon too
2005-11-29 16:40:43  <dhickey> great, thanks for your help and good luck with your work
2005-11-29 16:41:05  <Rudas> yeah, thanks. :-D
2005-11-29 16:41:11  <Rudas> Happy coding!
2005-11-29 16:41:21  <dhickey> you too