From: Jo Butler <pjgb@mrc-lmb.cam.ac.uk>
  To  : RASMB@BBRI.HARVARD.EDU
  Date: Tue, 15 Apr 97 13:49:46 +0000

Data formats/database

Dear RASMBers,

I am writing again, responding to John Philo's warning to "speak NOW!", as
I
think that John has slightly misinterpretted the consensus which I thought
was
emerging in the discussion.  I think he is wrong to believe that there is
'no
major objection to the idea of a binary format becoming the "primary"
format', I
had thought that the consensus was that ASCII would remain primary, with
the
option for binary available to those who needed it.  This seemed to be
the
import of Tom's mail of 07:57 on 10 April, with which I found myself
agreeing. 
Perhaps the ideal would be to have default settings which the user could
set
(preferably variable for different types of scan), so that he got the form
of
data he wanted with the minimum of fuss.  If this were combined with good
conversion programs, one could always obtain the other format, at a small
cost
in time, should one realise after saving the data that one wanted this.

Tom's suggestion of a database format for holding the data files could
certainly repay further consideration.  At present I find myself almost
automatically renaming and moving my data files, as the current Beckman
defaults, while excellent in avoiding any conflict and overwriting, are
not in
the least intuitive (at least to me) and I cannot imagine being happy to
sort
out what file is what at a late date.  The only way is to maintain a
comprehensive log book and it would be nice if this were more automatic on
the
machine.  Currently the Beckman software does not usually display the
"comment"
field when one is making calculations on the data, so this is much less
useful
than it might be.
What about a database in which the user is prompted to supply a run
number
(perhaps with a default which increments for each subsequent run) and an
overall
comment?  Individual cells could also have a comment field and programs
would
reflect these comments back to the user when the data is accessed.  There
would
be no need for the comments to be repeated in the data for each scan on
an
individual cell during the run, though clearly a header containing time,
w2t and
temperature would still be needed.

Best wishes,
Jo
Jo Butler, 
MRC LMB,
Cambridge, UK.
Tel:+44 1223 248011
FAX:+44 1223 213556
Email: pjgb@mrc-lmb.cam.ac.uk