[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Line Separators

• To: Multiple recipients of list <imgcif-l@bnl.gov>
• Subject: Re: Line Separators
• From: Andy Hammersley <hammersl@esrf.fr>
• Date: Mon, 18 Mar 1996 12:39:56 -0500 (EST)

Eric Fanchon writes:

> I think we should depart from CIF as little as possible. In other words, once
> header and data are split, the header file should be CIF compliant.
> (this was proposed by David Brown on 17 jan)

> Don't we agree on that ?  and if we do, doesn't this give an answer to the
> 'line separator' question ?

I think that I should emphasize that I (I hope "we") do agree on that.

However, the second question on the 'line separator' is unfortunately not
answered. Prior to splitting the header and the data, the whole file must
be generally considered as a binary file, and it is the manner in which
"lines" are stored within this binary file which needs to be defined. The
manner in which this is done could be independent of the "normal" methods for
storing ASCII text, but if we can use a method which helps people examine
the contents of their files with standard system tools e.g. editors, then
I think we should try to do this.

Peter Keller says:

> There are several public-domain or shareware PC-Windows editors around
> which handle Unix text perfectly well. Linux is likely to become more used
> in a scientific context, as the operating system for PC's, anyway.  In
> crystallography, both X-plor and CCP4 can now be run under Linux, and data
> processing software can't be far behind, as PC's become more powerful. 

Personnally I feel that the first sentence is a null argument. So what, if 
you can get special editors for DOS systems to handle Stream-lf text. The 
point should be to try to avoid people having to get specialist tools just 
to view the contents of their files. 

The second point is potentially more valid, but here we're getting 
into the realm of crystal ball gazing. Will DOS and DOS based systems 
never be of significant importance in the crystallography community, or 
at least that part of it which deals with image data ?  Certainly Linux 
is presently having a phenomenal success in the scientific community at 
present, but I still wonder, can it last ? I don't know the answers to 
these questions, but looking out at the huge world-wide market for PC 
DOS based systems I feel that it would be foolish to ignore them
completely.

Most commercial X-ray detector systems which I know about are PC/DOS
based at present, and I don't see this disappearing (even if at the ESRF we 
are presently replacing a number of PC-based CCD read-out systems with 
interfaces into Sun-workstations). 

So if Macintoshes use Stream-CR, what would happen if we tried to "view"
a Stream-LF file using a standard Mac editor ? And what would happen if 
we tried to view the DOS style file ? 

Unless, we are convinced that DOS-based systems are of no relevance for 
this image format, and won't be in the forseeable future, then I would
still recommend using the DOS-style \cr\lf line separators on the grounds 
that this will be the most interpretable on the most systems using native 
system tools.

-----------------

It's good to see that memory mapping has made it to some Un*xes. Does 
mmap() imply particular or minimum blocking sizes on particular systems ?
We should certainly try to leave sufficient flexibility to allow people
who want to, exploit memory mapping.

    Andy

Reply to: [list | sender only]

• Prev by Date: Re: Line Separators
• Next by Date: prototype CBF format
• Prev by thread: Re: Line Separators
• Next by thread: imageNCIF Convergence ?
• Index(es):
  • Date
  • Thread