Introduction
Attila Szegedi
my name is Attila Szegedi, and -- among other things -- I'm working on
a remote debugger architecture for Mozilla Rhino (JavaScript
implementation for the JVM). I did serve as the project maintainer for
Rhino for some time, and am still a committer to the project.
The background for my debugger involvement is that at my day job, we
have Rhino at the core of our mission-critical systems. This is not
Web and browser related, rather this is an embedding of a JS runtime
in a server-side system. Earlier this year I wrote a proprietary
remote debugger server and client that uses JSON over TCP as its wire
format and has the usual features - remote evaluation, watchpoints and
breakpoints, conditional breakpoints, step in/over/out, stack trace,
etc.
There was some interest expressed on Rhino mailing list in May about
having a "standard" implementation of a remote debugger for Rhino, so
I obtained my employer's permission to go open source with our
debugger. It's not done yet, although the work-in-progress is
accessible in a public Subversion repo [1]. Reason it's not done yet
is that I'm replacing our JSON wire format with Google V8 wire format.
Why V8? Well, more people pointed it out and specifically asked for
it, including Simon Kaegi (also on this list). I'm not particularly
attached to any wire protocol and would love to support one that has a
decent debugger client. From this point of view, Opera Dragonfly also
looks quite interesting, and Bespin endorsing a debugging wire
protocol would also be swell.
As you can see, my background and goals are a bit different -- I'm not
operating in the browser space, but rather in JS-on-server space.
Regardless, I'm interested in the discussions going on here,
especially since Bespin too is not targetted at developing code
running in browsers either, so I think we have a lot in common.
My interest is in having the community agree on a wire protocol (and
transport method) for remote JS debuggers[2]. My vote is on JSON over
TCP, with components running JS exposing a listening server socket and
debuggers connecting to the socket (rather than the other way round).
[1] Pointers to my current effort and description of its current state
can be found here: <http://groups.google.com/group/mozilla.dev.tech.js-engine.rhino/msg/42881aacd5974765
>
[2] Actually, I have some more fundamental motives -- I'd like to
have a JSON/TCP protocol that to the extent possible is language
independent; not suitable for JS only, but one that could be
reasonably used to additionally debug at least Ruby and Python
programs as well. Again, I think this is in alignment with Bespin goals.
Attila.
johnjbarton
On Jul 3, 11:28 pm, Attila Szegedi <szege...@gmail.com> wrote:
> Hi folks,
>
> my name is Attila Szegedi, and -- among other things -- I'm working on
did not notice the email notifications. I get about 20 a day from the
Firebug group, 2/3 are spam. So often I use the link in the first one
to navigate to the management page then moderate them all. Then I
delete all of them email from Google Groups at once. Again, sorry.
jjb