Results of the McStas users survey
The McStas users survey was ended on September 23rd. This page
presents the main conclusions drawn from the survey.
A total of 22 individuals participated in the survey.
Platform - summary
The platforms (OS/architecture) on which McStas is run is (in order of
Linux and Windows on PC hardware are the most important
platforms, more or less as expected.
- Linux / i386 (21)
- Windows / i386 (9)
- IRIX / MIPS (3)
- Linux / alpha (2)
- HP-UX / PA-RISC (1)
- Digital Unix / alpha (1)
- MacOS 9.? / PowerPC (1)
McStas version - summary
The preffered versions of McStas are (in order of appearance).
seen that the McStas users are roughly divided into three large groups using
versions 1.4, 1.5 and 1.6-ill. Two sources of error are important to
emphasize here, one beeing the relatively low number of survey
participants, and the other that McStas v. 1.5 actually returns "1.4"
when run with the -v switch. So if the responding users used "mcstas
-v" to determine their local running mcstas version, some of the 1.4
users could actually be 1.5 users.
- McStas 1.4 series (7)
- McStas 1.5 series (7)
- McStas 1.6-ill series (7)
- Version previous to 1.4 (1)
Most prominent problem - summary
The most problematic area for McStas according to the users is (in
order of appearance, disregarding the default choice "No special area").
The order above corresponds nicely to what the developers have decided to
emphasize on for the next release. Most likely, the current
perl/perlTK/PDL/PGPLOT setup will be replaced for the next
release. Support should also be better now, since a full-time McStas
administrator is now employed at RISØ, general agreement
between the McStas developers has been achieved and since the mailing list is
functioning again. Lacking stability / consitency in the output of
McStas is also beeing worked on, it has been decided to include test
procedures for checking performance of new McStas kernels as well as
new components. The lacking feature mentioned above regarded the
possibility of including a Laue-type time-of-flight area detector. We will work
on this. Regarding the reported problem of the current installation
procedure, it is possible that this answer regards the "total"
installation procedure, including support applications such as
PGPLOT. For now at least, it has been decided to stick with source
packages only, keeping the current approach (./configure; make; make
install), which seems not to cause too many problems.
- Support applications (mcgui.pl etc.) (7)
- Lacking support (3)
- Lacking stability (2)
- Lacking features (1)
- Installation procedure (1)
Apart from the above input, many other relevant comments were given. A lot
of the comments supported the input summarized above, such as "replace the
current gui", other are very much in agreement with thought among the
McStas developers, e.g. the possibility to include an "adaptive
source", a larger number of included example instruments with comments
(also relevant for test purposes, cf. above). One user reported the
need of a sample data file for Be to supplement the current BeO data
file. If someone has such a Be.dat file, we will be happy to include
that in the next release. One user mentioned the possibility of
including a "step-by-step" installation document for different
systems, we can certainly include such a document for the mostly used
platforms, and will be happy to include information from people
running McStas on other platforms.
The results of the survey is largely in full agreement with the
decisions taken for the next McStas release, which is very
satisfying. Hopefully, the problems that people are currently pointing
out will be sorted out by the next release of McStas. A hard deadline
has not been set yet, but the developer team is aiming at releasing
next McStas version before 2003.
RISØ, September 30th 2002