Re: Good debugger?

From: Roy Terry (royterry_at_earthlink.net)
Date: 11/17/03


Date: Mon, 17 Nov 2003 14:58:10 GMT

Bgc wrote:
>
> Cameron:
>
> Stepping through a script sequence, one line at a time, and watching
> the variables change values as you do so is a much better approach to
> learning a language, in my opinion, then inputing values to a function
> and observing the output. The "closeness" that you get to the code
> (or whatever word that you want to use) is acheived by stepping
> through it with a good debugger. What I really like about RamDebugger
> is that you change values even while the code is executing (I don't
> think I've ever seen a debugger do that before).

Sure, that's true. And yet what's not mentioned is the fact that
you already had a debugger mindset and so did not need to work through
the concept of one program starting and stopping another which is
big conceptual magic not to mention the potential hassle, illustrated
in this thread, of getting the debugger happy with your environment and
getting it to ingest your source code. (And then there is the
vexing possibility that the debugger will subtly interfere with
code behavior (or that the user will get confused into suspecting
same)).

But on the question of stepping through the code while learning,
here's what's not mentioned: Every Tcl console (and
especially Tkcon) is a single step tool. I suspect
those of us unmotivated to use debuggers just run things at the
console a command at a time and see what happens. For example,
if I'm writing a complex RE, I just copy/paste to tkcon and
can very easily get all the iterations and trials I need before
pasting the result back into the source file. So if you say you
have ten lines in a row that are all mysterious and so must
be debugged, I will say yes happens to me too but not often
enough to want to switch to running a separate tool. However,
hearing the good words about RamDebugger makes me interested
in trying it if I ever do get that stuck.

I still think the best *general* advice to newbies is to
get comfortable with Tkcon.

Cheers,
Roy Terry
(Now debugger free for 5 years - :)

> > .
> > .
> > RamDebugger is indeed a significant achievement. It intrigues
> > me, though, that you regard a debugger as "a must for newbies".
> > That's not the way *I*'ve taught Tcl to beginners. What am I
> > depriving them? How would lack of a debugger have impaired
> > your Tcl learning?



Relevant Pages

  • Re: Learning C
    ... I stress that I talking about "stepping through an application". ... And how can stepping through an app be bad? ... I did not step through the code with a debugger. ... "examine", to me, means to study the source code. ...
    (comp.lang.c)
  • Re: Learning C
    ... with a debugger. ... I stress that I talking about "stepping through an application". ... But often loggers affect the app. ... "examine", to me, means to study the source code. ...
    (comp.lang.c)
  • RE: Step by Step skips large chunks of code
    ... Putting on SP1 the problem occurred. ... the debugger seems to work OK in the main thread ... let us monitor the threads in the Threads window. ... Then start stepping code. ...
    (microsoft.public.vsnet.debugging)
  • Changing source on the fly (Was: Lisp Book(s) in the K&R Style)
    ... Is it possible in CMUCL to change the source when stepping through the ... the source of the very function or loop you are ... curious when I will find it in any lisp debugger. ...
    (comp.lang.lisp)
  • Re: managed c++ and native c++ interop
    ... There are know bugs in the debugger interop stepping code. ...
    (microsoft.public.dotnet.languages.vc)