#9 wishes...

open
nobody
None
5
2003-02-03
2003-02-03
James Mc Parlane
No

It would be a "Very Good Thing" if the introspector could
also be pointed at the following types of files and have
an XML interface file extracted.

1) lib.so files
2) lib.a files
3) app.java java 'binaries'
4) app.exe C# 'binaries'

Sound crazy.. but I could real use this. 3 an 4 should be
possible by reflection.

I think 1 and 2 are impossible because you can't get
function arguments out of a lib file, but I could be wrong..

Discussion

  • Logged In: NO

    lib.so .a files can be first extracted with nm and c++ filter.
    nm /usr/lib/libstdc++-3-libc6.2-2-2.10.0.a | c++filt

    app.java you may be ables some patches to pnet or java dumpers .
    I would patch the jit and the jvm to dump out the needed info.

    same for pnet

    I will do the NM one first. I need that myself.

    mike

     
  • Logged In: YES
    user_id=147871
    Originator: NO

    sorry for the late response.
    I have an the beginnings of an objdump introspector running.
    will publish that soon, I have now an xml patch for objdump, not yet rdf.

    first we need to discuss the idea of an owl:class in a owl:ontology and how that relates to a program.
    this is going to get very deep into my personal and internal representation of data.
    the idea is to add a rdf predicate interface to the objdump printf statement.
    first capturing the output.
    The printf statements are each events in the runtime of the objdump program.
    each one of them is in fact a class definition on its own right.
    That means that each statement in a program is really a class that is instanciated when that statement is executed.
    The instance of a statement occurs when the statements generated code is executed.
    That means that in effect, each block of executable code is traceable back to some outside (or internally synthesied) source.
    thus we define a class associated to the instruction pointer of a process.
    The instance of that instruction contains other registers and the associated data, we can call the data context.
    So a running program can be seen as a sequence of actions by actors. the theme of the program is determined by both the data in the program and the data from the world around it. The entire set of inputs and outputs of the program are give only in a point of time in the real world.

    More to come.
    mike