[IPython-dev] Comments about IPython.frontend, getting ready for merge of ipython-sync-frontend

Barry Wark barrywark@gmail....
Thu Aug 7 18:36:52 CDT 2008


Brian et al.,

I don't have time to give your comments the time they deserve tonight
so I wanted to just weigh in quickly on merging Gael's branch before
the upcoming release. There are some significant changes I want to
make to the frontendbase interface, mainly making the concept of a
execution block more explicit, but I realize Gael has made great
progress with his branch that diverged before I made those changes. I
think the synchronous frontend should be merged in for the release if
possible. I will hold of on my changes until the release has settled
and we have time to evaluate how the newer frontend architecture. I
have just one other comment, in line below, for the moment and will
try to get you a more proper response soon.

On Thu, Aug 7, 2008 at 2:11 PM, Brian Granger <ellisonbg.net@gmail.com> wrote:
> Hello all (especially Barry and Gael),
>
> We would like to merge the ipython-sync-frontend branch into trunk
> ASAP.  Fernando will be sending out an email later today describing
> our timeframe for the upcoming release.  Meanwhile, I wanted to
> provide some informal code review and feedback on the stuff in
> IPython.frontend.  I haven't looked over the wx or cocoa specific
> stuff, just the general stuff.
>
> Overall, I am _really_ excited about this code.  Many thanks to Barry
> and Gael for pushing this stuff forward and making the first go and
> using the new core.  I talked to Fernando last night and both of us
> are completely committed to getting this stuff into the next IPython
> release, which is coming soon.  I should preface my comments by saying
> that I don't think all of the comments need to be address before the
> merge or the release.  Some of the comments are more long term.  Also,
> Fernando is going to look over the new stuff in IPython.kernel.core (I
> haven't looked at this stuff).
>
>
> Here are my comments:
>
> ===============================
> Comments about IPython.frontend
> ===============================
>
> General
> =======
>
> We need a robust way for modules in the frontend to handle interfaces
> if zope.interface is not installed.  Currently each module tries to
> import zope.interface and then builds a mock zope.interface, like
> this::
>
>        try:
>            from zope.interface import Interface, Attribute, implements, classProvides
>        except ImportError:
>            Interface = object
>            def Attribute(name, doc): pass
>            def implements(interface): pass
>            def classProvides(interface): pass
>
> This code should probably be put into a common location, something
> like frontend.zopeinterface, so that it can be reused.  But eventually
> when
> IPython.core is created, this should probably be moved there.
>
> The docstrings are vague and outdated at points.  More detailed and
> updated docstrings would be helpful.  Also, all docstrings should use
> the epydoc+ReST format.
>
> IPython.kernel.core should probably be moved to IPython.core.
>
> Once Fernando's testing branch has been merged, more tests should be written.
>
> frontendbase.py
> ===============
>
> Configuration should be done using the approach in IPython.config.
> See IPython.kernel.config and IPython.kernel.scripts.ipcontroller for
> examples.
>
> IFrontEnd has references to Twisted in its docstrings.  Because the
> basic frontend should work without Twisted, the docstrings should be
> updated to clarify how the interface plays with Twisted, even though
> it is optional.  We really want to document for developers that the
> base frontend is synchronous and doesn't require Twisted, but that the
> interface has a particular design that allows it to be wrapped into an
> asynchronous frontend (like some methods pass through the result).
>
> The design of the frontend seems good though.  We might eventually
> want to think about using the notification stuff in the frontend
> though.
>
> Why does FrontEndBase not have a complete method?  Other frontend
> classes do have this method and its seems like the base class should
> have it too.  Also, there are oher methods in the interpreter that we
> might want to propagate up to the frontend, like push and pull.
> Thoughts?  If these additional methods go in, there should be separate
> asynchronous versions that return deferreds, but these would go into
> asyncfrontend.py.
>
> asyncfrontendbase.py
> ====================
>
> Because this frontend us designed for use with Twisted, I don't think
> we need to protect the zope.interface and Twisted imports.  Also setting
> Failure to Exception will break things as they don't have the same interface.
>
> Design looks good though.
>
> linefrontendbase.py
> ===================
>
> This looks good as it is simply a concrete subclass of FrontEndBase.
> General comments above about docstrings apply.
>
> prefilterfrontend.py
> ====================
>
> It is fantastic that we can combine ipython0 and IPython.kernel.core
> code like this.  This is a great way of using the new stuff, but
> getting the more complete feature set of IPython.*  But I think this
> type of thing should be done in the lower-level Interpreter, rather
> than the frontend.  This is for a couple of reasons:
>
>        1. This would give all users of the Interpreter access to completion and the
>           prefilter stuff.  We really need this stuff in the parallel
> computing stuff,
>           as well as non-line-oriented frontends.
>        2. The Interpreter already has some of this logic and it is simply being
>           repeated.
>        3. The current implementation requires that the Interpreter is in-process.
>           For connecting to a remote Interpreter, this won't work.
>
> But, we might run into a problem with doing this stuff (that cool hack
> that is begin used currently) in the Interpreter.  The reason is that
> the Interpreter is block based, whereas the prefilter stuff in IPython
> is only line based.  Thus, doing this refactoring might be a little
> more involved.  But it is the eventual design we want to go for.  This
> stuff probably won't be done before the merge. We can work on this at
> SciPy.

I think we should, as Gael points out, think carefully about design
decisions early on. Namely, is there a philosophy about what belongs
in the core, what in the Interpreter itself, and what in the, e.g.
frontend? I tend to think that the Interpreter should be *just* an
enhanced python execution engine, knowing as little as possible about
where input comes from or where it goes. Even the terminal frontend is
a frontend. Similarly, the more we can separate the concerns of
input/output from the Interpreter the easier it is to move things
between processes/hosts/etc. later on. I realize there is a tension
between those who want a lean, fast, Interpreter and those who want an
abstracted, layered architecture. I obviously don't have the solution,
but hope that in the coming days/weeks we could all articulate a
philosophy or blueprint for where we want things to end up.

>
>
> Great work everyone!!!

Thanks. Cheers,
Barry

>
> Brian
> _______________________________________________
> IPython-dev mailing list
> IPython-dev@scipy.org
> http://lists.ipython.scipy.org/mailman/listinfo/ipython-dev
>


More information about the IPython-dev mailing list