Dear BUMmers,
I'm pleased that Jim has brought this issue out into the open. It was
actually first mentioned a year ago when the Linux version of XWin-NMR was
announced. I seldom use XWin-plot and, as there was no danger of my getting
money for either the new software or a new computer to run it on any time
soon, I just kept (relatively) quiet. I've had cause to use XWin-plot a
couple of times, solely for it's ability to export a plot in a graphics
format. In the past, a couple of Bruker chaps have been in and tried to
show me how to use it. After these experiences, I think the replacement of
edg/plot with XWin-plot will, for my service, result in a major loss of
speed, flexibility, precision and overall usefulness.
XWin-plot is possibly the worst graphics program (let alone the worst NMR
program) I've ever had the misfortune to use. I found the whole scheme of
operation counter-intuitive, slow, clunky, difficult, unforgiving and
basically a complete pain. The user interface is a total muddle. No thought
appears to have been put into the ergonomics of using the program. The help
information is pathetic, and so is the printed manual. The program possibly
contains most, if not all the functionality that is required, but getting
at it quickly and easily, let alone controlling it, is seriously difficult.
Both Xwin-Plot and XWin-NMR are typical engineers' products. Functions are
scattered more or less at random. The highly varied working practices at
customers' installations do not appear to have been investigated, so have
not been allowed for.
I think the basic premise of the program's operation is wrong. According to
Mike Engelhardt, it's meant to be like other graphics programs. When
pressed, he said he'd modelled its operations on PowerPoint, and no, he'd
never used Photoshop. However, as others have already mentioned, NMR
spectra are usually _working_ documents. An NMR spectrum produced in a
service environment for use by chemists in the lab is _not_ a graphics
exercise - it is a piece of CAD work. People take measurements from them
using a ruler, hold them up to the light to overlay them and compare
things, transfer information from one spectrum to another. I teach this
kind of stuff, and I do it myself. Precision and consistency in plotting
are required to do this. The edg/plot system fulfils this to a much greater
extent than XWin-plot does.
One of the basic ideas of XWin-plot is to drag an object to where you want
it with the mouse. This is display, or presentation _not_ CAD. The position
and size of the 'objects' in the plots I produce were worked out by me with
a ruler and are positioned to the millimetre. edg/plot allows me to do this
positioning, and retain it for the future. I think the only thing I might
adjust by eye on the final plot is the peak height, CY.
At the Users' Meeting last week many people spoke of a 'steep learning
curve' for XWin-plot. This can and should be engineered out, by better
design. It's long been known that many software engineers like a steep
learning curve, just to keep the riff-raff out of their systems. Fair
enough perhaps with monsters like command-line Unix, but XWin-plot is a
program that many people must learn to use, and quickly.
The design of Graphical User Interfaces has been the subject of serious
academic study for decades now. We have a significant research interest in
the subject here. Programs and systems in a competitive market have had
their GUIs honed to match users requirements, and there is good knowledge
in the industry of best practice in GUI design. Bruker don't appear to have
noticed.
A recognised, and recommended method of producing a good GUI design is to
pay attention to what difficulties users, especially novice users, have
with the program. As a designer, your starting point must be that whatever
you have dreamed up, no matter how cool you think it is, may not suit all
(or any) users. You must accept the need for change, and write your
software in such a way that the interface _can_ be changed. You then get a
novice user, one who knows what s/he wants to achieve, but hasn't used the
new program yet. Then, while one person _gently_ guides the user, another
sits in the background taking notes about where the problems arise. You
then alter the program to accommodate the user's difficulties, and try
again - repeatedly.
I do not believe the Bruker have ever used this method for any of their
programs. I think it is about time they did, and I volunteer to be the user.
OK, this rant is long enough. Let's hope we hear from all the concerned
participants at last week's meeting, it might do some good.
Alan
--•--------------------------------------------------• Dr A.S.F.Boyd Chemistry Department NMR Spectroscopist Heriot-Watt University phone: +44-131-451 3214 Riccarton fax: +44-131-451 3180 Edinburgh email: a.s.f.boyd@hw.ac.uk EH14 4AS http://dava.che.hw.ac.uk/ Scotland •------------------------------------Made on a Mac-•
This archive was generated by hypermail 2b29 : Tue Dec 17 2002 - 01:04:24 PST