--- Akim Demaille <[EMAIL PROTECTED]> wrote: > >>>>> "James" == James Michael DuPont <[EMAIL PROTECTED]> writes: > > James> In the long term, There is no reason to use files to transfer > James> this information between programs, we should make a > James> visualization module and gui for gcc. > > Well, in the short term, I do plan to have Bison produce more graphs. > For instance, Bison could, in addition to YYDEBUG, support some form > of YYVCG_PARSE_TREE which outputs the parse tree. I don't plan to > have Bison parser equipped with means to call vcg!!!
Sure, I have created a bison.xml that replaces bison.simple for dumping the parse tree and stack to xml. But in the long term I do plan on linking bison in. Bison, bash, m4, gcc, all types of tools. I plan on creating a libintrospector that allows you to link in all types of gnu tools for data visualization. Of course it wont happen over night, and untill then xml will help in the storage, it is easier to parse and easy to generate. > James> My plan is to load the GCC and VCG directly as modules into > the > James> introspector and extract whatever data is needed when it is > James> needed. > > James> In my opinion there is no real reason to create intermediate > James> files at all with AST information, > > AST != parse tree. Sure, in the long term there is no reason to create intermediate files at all. It might sound crazy and radical, but we are talking about structured data transport here. These files are just serving as buffers of data between programs. We need to be able to eliminate the files and link the needed tools togeather... a form of secure data layer. mike ===== James Michael DuPont http://introspector.sourceforge.net/ __________________________________________________ Do you Yahoo!? Faith Hill - Exclusive Performances, Videos & More http://faith.yahoo.com
