BASIN client on macs (and other issues)

Dear Bason-ites,

I'm pleased to say that since we published our CISE paper on the
preprint server, our hits on the webpage have increased dramatically,
and we've gotten a few external emails from people who seem genuinely
interested in using BASIN. In fact, as we speak, I am starting to use
BASIN for my own science -- analyzing cluster-scale N-body simulations
for use with lensing reconstructions.

I have hit a couple of snags, especially when using a client running
Leopard, including:

1) (Not Mac specific) I am unable to connect to a VisIt session from
behind a router. Even after port-forwarding 22, and 5600-5609, I can't
get this to work. I should not that I CAN connect to a "sim" from my
office machine, and thus the problem isn't strictly speaking on the
VisIt side. Finally, while the instructions for building visit are
fairly complete, there are few instructions for configuring VisIt (e.g.
setting up the "hosts", using the "-dir" flag, etc.) for someone using
BASIN as a remote client.

2) I cannot do the following on either my home or office Macs:
    a) "Open a file" using the dialog in the client. (Though I can
easily open a file with the reg=Region"myfile" command).
       The error in my log appears as:
/System/Library/Frameworks/Python.framework/Versions/2.5/Extras/lib/python/twist
ed/python/failure.py in raiseException(self)
    256 information if available.
    257 """
--> 258 raise self.type, self.value, self.tb
    259
    260

NameError: name 'filesystem' is not defined

    b) I cannot drop and drag anything, and can only double-click on
variable names.
   
3) gnuplot does not work (at least on my office Leopard box), since
presumably it ssh's with a -Y flag, while Leopard requires -X in order
to do x-forwarding.

4) MatPlotLib doesn't work, since for some reason, numpy won't install
via fink for Python2.5. I could, of course, switch to 2.4, which should
presumably take care of the problem.

5) Save Batch (and presumably Load Batch) doesn't work. I get the error
in my log:
***************************************************************************
Exception in the IPython Engine.

Engine action that caused the error:

engine: 0
method: execute(lines)
lines = buffer_history.read ()

Full traceback:
---------------------------------------------------------------------------
NameError Traceback (most recent call last)

/home/goldberg/<string> in <module>()

NameError: name 'buffer_history' is not defined
***************************************************************************
       
***************************************************************************

------

Those are the main errors I've encountered so far, but there are some
more general meta issues which need to be dealt with including:

1) I would like to write a new i/o package to generate initial
conditions for Gadget simulations and to read them in. While I've been
moderately successful in the past trying to parse the I/O package, a
document on the Wiki instructing me on writing new datatypes (would be
most useful). Who has done the most work on this the most recently?

2) I would like to add a few additional menu items to the GUI. James, I
know your time is quite scarce, but if you could give some pointers on
how one might go about that, I'd be most grateful.

3) The API is continuously out of date, and we need lots more examples
on how to do things.

-----

Now, some of these are general statements, and as such, I expect that
they won't be dealt with. That needs to change. It will be a major
topic for discussion at the Wednesday meeting. In the meanwhile, if
anybody wants to get a jump on this by giving specific suggestions, I
would be most obliged.

Sincerely,

Dave
Received on Mon Jun 2 14:52:27 2008