iNMR news #86

[GB, author of iNMR]

Today we start the public testing of iNMR version 5.
The exact name of today's version is 4.9.0. It is not a substitute of
current version 4.2.
Version 4.2 is complete and correct. It is still living and still
growing.
Version 4.2.1 will appear soon.
Version 4.9 is unfinished and full of bugs.
It is a complete rewrite of iNMR in 64 bits code, suitable for today's
and tomorrow's computers. When complete and corrected, it will take
the definitive name "5".

The program is far from being finished. Many important and large
components are completely missing, like:

- The Overlay Manager;
- All Simulations;
- Line Fitting;
- The Scripts Interpreter;
- The Search Module;
- The Plug-ins;
- The Cross-Peaks Manager;
- DOSY Processing;
- The J Manager;
- Container documents.

A lot of things are however present. At first sight it's the same old
program you already know. At a second sight you will discover many
subtle changes, which are supposed to be improvements.
The purpose of the public testing is dual:

1. We want to find and correct all the errors still present.
2. We want to hear your opinions about the new interface.

Any recipient of this newsletter can be a volunteer. Those who will
subscribe to the iNMR group today or later are excluded.
Download the program from the address:

http://www.inmr.net/downloads/5_0.zip

Install it on a single Mac and ask for the unlocking key. Explore the
program for as long as you can. If you discover an error AND if the
error can be reproduced, send an email to in...@inmr.net.
If you want to discuss about the new interface and suggest something
different, publish your comments here to the group.

DANGER !
It is almost sure that INMR5 can damage some of your old iNMR files.
Do not work with your original files, but with copies.

Here is what the READ ME file says :

This program is under construction. Many important parts are missing.
The program can contain many errors.
This program might ruin your data and might produce wrong results. You
are warned not use this program for your study or work.

Hardware Requirements: Intel Core 2 Duo processor or higher.
Software Requirements: Mac OS 10.6 or higher.

Check for Updates! When running this program, periodically check if
updates are available with the command:
iNMR > Check for Updates

Expiration of the license
If you obtained a license for this program, it will expire before the
release of the definitive version.
The license might expire well before that date if you forget to send
you feedback.

How to obtain a key to unlock this program on a SINGLE computer:

1) You must be a member of the iNMR group at http://groups.google.com/group/inmr
since September 2011 or before.
2) Install iNMR5 on your Mac, run the program, open the dialog iNMR >
Registration.
3) The user name must coincide with your email address (see below).
4) Follow the instructions to send an email
5) When sending the email, use the same address you have used when
joining the iNMR group.
6) You can't ask for more than one key.

Remember to send your feedback to in...@inmr.net or to the iNMR group.

[Pascal Fricke]

Dear Giuseppe,

many thanks for pushing out the iNMR 5 beta and for letting us see a
first glimpse of the upcoming improvements of the next major release
of iNMR.

It is in my opinion a very good decision to change the underlying code
to be 64bit compatible and to use the latest Mac OS X APIs. Since even
iNMR 4.2 runs so smooth and fast, I cannot "feel" any speed
improvements, but did you do any benchmarks to find out if the
processing speed has improved?

What I do like very much about the changes in the frontend is the use
of the "black theme" (e.g. in the monitor window or the phase
correction window). It might sound naive, but this makes iNMR look
more modern and underlines its nice integration into Mac OS X;
furthermore, it now looks more similar to many recently published
applications, especially via the App Store.

But I would drive the whole thing even a little further and use this
theme for *all* hovering controls. Especially the palette, because the
design looks now a little inconsistent.

To be honest, I don't like the new design of some buttons. This is
especially true for the design of the buttons of the palette and some
buttons in the "Preferences" dialog ("Factory Defaults", "My
defaults", "Choose"...). Why are they of a darker grey and different
shape than the other buttons? To me, this somehow looks a little old-
fashioned and does not convey the spirit of the new Mac OS X design.

I like the introduction of new cursor icons for selections, the old
ones always looked a little "general". But I think this should not be
accompanied by the drawback that precise selections are not longer
possible. This especially applies to the Cutter, Interpolator and
Monitor. These cursor icons don't have a visible "hotspot" (like for
example a cross) which allows to precisely set the point of selection;
so the user has to "guess" where the initial selection will happen.


So far, this is my very personal judgement. I hope my criticism does
not sound too harsh, it really is not meant to be.

Here I have a link to a Mac OS X program (Pixelmator) which is very
well integrated in Mac OS X and makes extensive use of the newly
introduced dark controls. I like the frontend very much and maybe you
can be a little bit inspired by the design (especially of the Tools
palette), as you can see on the screenshots:

http://www.pixelmator.com/learn/2010/11/bokeh-effect/

With a comparable design iNMR might look even more like a genuine Mac
application (see for example Final Cut Pro X), although conservative
people might call it "too fancy".


Kindest regards,

Pascal

[GB, author of iNMR]

In this moment I more concerned with functionality and less concerned
with appearance, for a practical reason. I know that any new program
contains many errors and it will take time to discover all of them.
The problem is very serious in our case, because the user base is very
small. There are thousands more beta testers for a program like
Firefox than iNMR. To reach the same level of accuracy we need, in
theory, to multiply by 1000 the length of the testing period, which is
obviously impossible. The only thing I can do is to start testing the
functionality ASAP.
Cursors and windows can be changed at any time, because they require
much less testing. I had already decided that I will draw new cursors
for the definitive release.
Personally I like black windows, but there are some problems. Normal
windows are more flexible and text on dark windows can be difficult to
read. Take for example the palette. The standard black panel (in Snow
Leopard, which is what I am targeting at this moment) only contains
the close button, not the zoom button, which is quite important in our
case (it lets you change the layout of the palette).
Certainly we should aim at more consistency. Version 4.9.0 contains a
variety of buttons. It would be probably better to have a single kind
of buttons in all windows, but I am not sure. I use classic buttons
for "OK" and "Cancel", because they are more visible, grey buttons for
less important functions. I find them more modern, so my impression is
very different from yours.
I have not tested the speed of the new code. There is no reason for
doing it: the interface has changed, but the engine remains the same.
Numbers are still stored as 32-bits, not 64. It was well known, 20
years ago, that programs like iNMR were faster and more accurate if
they used 64-bits floats. RAM was the limiting factor then and now
too. When working with 1-D and 2-D spectra iNMR is already fast
enough, so we don't need bothering. Where iNMR is slow is in the
transposition of large 3-D matrices. This is both a problem of RAM and
of data-bus. Moving completely to 64-bits would make things worse.
64-bits precision was already possible (and welcome) with the PowerPC
15 years ago. iNMR has always implemented it for critical numerical
tasks, like Linear Prediction, matrix diagonalization, etc.. With
version 5 we have added 64-bits addressing, which is a completely
different thing; a useless thing, in the NMR field at least, but it's
going to become a requisite, soon or later, so we want to be prepared.
Rewriting the interface is already a daunting task: there is no time
to work on more things. For example, it is very intriguing to use
OpenCL for the FT…
Although the new version is just as fast as the old one, it does many
more things. Try, for example, to drag the margins of a plot. The
behavior is very different. Version 4 looks very lazy, when compared
with version 5. I can make many similar examples, but you can easily
discover them by yourself.