Dear All,
This is in reply to Rainer Haessners Mail:
> in the past we had an interesting problem. We use our spectrometers
for data acquisition only
> and process the data at diffent stations. The data itself become
copied using the rcp -r
> command. Processing of spectra is no problem. But some users were not
able to produce
> correct plots of their spectra. They produced something like the frame
of the desired plotout
> (scales, title, peak picking) but the spectrum itself was missing. The
plot preview always
> showed the correct spectrum.
> The solution of this strange behaviour had nothing to do with XWIN-NMR
in a direct way.
> That users, which couldn't produce correct plots had the line
>
> umask 027
>
> in their .cshrc all others had
>
> umask 022
>
> Of course umask 027 is somewhat more secure. But using this option you
produce a lot of
> possible trouble (not only that plotting problem). For instance the
simple command
>
> lp <file>
>
> wouldn't work.
> Hope this helps others to avoid this pitfall and maybe the Bruker
software engineers should fix
> this behaviour, if possible (i believe, there must be a program with
the suid flag responsible for
> this behaviour).
The story is as follows.
Under IRIX5.2, the lp command would not print files where the user lp
had no read permission. You would get the error message :
Cannot access filename
request not accepted
>From IRIX5.3 on (maybe with patches available from SGI) this behaviour
was changed. If the user lp has no access to the file, the file is
copied to another location (where lp has write permission) and the file
is made readable for lp. Now the file can be printed. So much for the lp
side of things.
In XWIN-NMR, we don't actually send one complete file to the lp queue
with all the information in it. Instead, we send a so-called "meta" file
with references ("links") to the actual files. This saves us the copy
procedure, which is (or was, at least) important for biggish 1D or 2D
spectra. There were space problems on the old, small disks etc.
The only thing we can do to fix this problem is to restructure the whole
plot mechanism to adopt the new, intrinsic, "copy on demand" feature of
the more recent lp command versions. Alternativley, you can use
XWIN-PLOT to generate the plot. XWIN-PLOT doesn't have this problem, it
generates one file and you can even specify the -c option on the
XWIN-PLOT lp command line. Or you give at least read access to your
data, then there is no problem anyway.
Best wishes, Mike !
===================================================================
... ... Dr. Michael Engelhardt
. * . . Software Department
. . . . Bruker Analytik GmbH
B R U K E R Silberstreifen
. . . . D-76287 Rheinstetten
. . . . Voice: +49-721-5161-440
... ... Fax: +49-721-5161-480
Email: michael.engelhardt@bruker.de
===================================================================
This archive was generated by hypermail 2b29 : Sun Dec 31 2000 - 01:15:03 PST