NSCLSpecTcl is a low overhead dynamic histograming framework for Nuclear Sciecne experimental data. It was written for the National Superconducting Cyclotron Laboratory (NSCL) at Michigan State University. When the NSCL experimental program finished and the NSCL was succeeded by the Facility for Rare Isotope Beams (FRIB), the FRIB adopted NSCLSpecTcl for use in online and early offline analysis.
Note that NSCLSpecTcl should not be confised with the Tcl GUI interface generated named SpecTcl. In this document the word ```SpecTcl`` will be used for brevity to mean NSCLSpecTcl
NSCLSpecTcl development is now hosted at:
http://github.com/FRIBDAQ/SpecTcl
The documentation for NSCLSpecTcl is at:
Various documents describing SpecTcl are in the section labeled "What is SpecTcl?" on that web page.
SpecTcl is developed under Debian Linux and has several dependencies, some are built in, some are expected to be installed by the system owner. As the FRIB actively develops NSCLSPecTcl, it is possible that the list of dependencies below are incomplete.
One can install NSCLSpecTcl in three ways. These are listed in decreasing order of our preference for you;
- By using one of the FRIB supported singularity containers and filesystems (see https://docs.nscl.msu.edu/daq/newsite/DAQFromContainers.pdf for a desscription of our container systems and how to install them).
- By building from a distribution tarball
- By building from a clone of the repository.
https://github.com/FRIBDAQ/SpecTcl/releases
Provides a set of the releases we would prefer you to use. Each of those releases has several 'assets' or files; If a release is named x.y-zzz; it will have the files:
- spectcl-x.y-zzz-{container}.tar.gz Where {container} is the Debian release for one of our containers. These are binary distributions which should be installed in the directory tree that is bound to /usr/opt from our containers and filesystems in general it will have a single directory tree named x.y.zzz which should be unrolled into a directory that will be bound to /usr/opt/spectcl (e.g. /usr/opt/spectcl/x.y.zzz).
- spectcl-x.y-zzz-dist.tar.gz This is a source distribution generated
by the
make dist, see however DEPENDENCIES below. if you want to build using this. - spectcl-x.y-zzz.{zip,tar.gz} these are labeled as "Source code (zip): and "Source code (.tar.gz)" respectivel in the release asset list. These cotain clones of the git repository tag that generated the release.
This installation requires the least knowledge -- once you have your containerized system up and running. Suppose you downloaded version 5.13-010's binary tarball for e.g. the bullseye container. You would have gottan a file named:
spectcl-5.13-010-bullseye.tar.gz
Hand that file to a user that is allowed to write into the directory tree that is bound
into the bullseye container's /usr/opt directory tree. Suppose they put that file
in their home directory and a script or alias named bullseye has been defined
to start the container with that directory tree bound to /usr/opt. Here's how to
install that new SpecTcl Version:
cd /usr/opt/spectcl # if necessary make this directory.
tar xzf ~/spectcl-5.13-010-bullseye.tar.gzDone, you now have NSCLSpecTcl version 5.13-010 installed for use in your containerized buster system.
Unwrap the git clones and cd into the main subdirectory of the resulting tree.
SpecTcl depends on a few other FRIB/NSCL products. Rather than make you build and install them, we have three scripts that allow you to incorporate them into the SpecTcl build. Each script is quite short but you'll want to read the comments that descdribe the most recently used version/tag of that product.
You will need to have git installed on your system to incorpoprate these.
- The Unified format library provides code that can decode NSCLDAQ data (NSCLDAQ is the data acquisition system adopted by the FRIB it too was developed at the NSCL). Suppose you see the command in the script unifiedfmt-incorp.sh
# Most recently used version 1.2
run the script as:
./unifiedfmt-incorp.sh 1.2to incorporate that version of the unified format verison.
- tcl++ Is a C++ wrapping of the Tcl API libraries. SpecTcl uses this library extensively. It's incorporation script is tcl++incorp Suppose you see the comment:
# e.g. ./tcl++incorp libtclplus-v4.3-001
# last used in this branch.
#
towards the top of the script. In that case, invoke it as:
./tcl++incorp libtclplus-v4.3-001- CutiePie is a displayer which allows you to view the spectra created by SpecTcl during its analysis of your data. It's incorporation script is qtpi-incorp.sh If you see a comment like:
# last tag was v1.4
Run this script as folllows:
./qtpi-incorp.sh v1.4Now all of the internal dependencies are incorporated and you can build SpecTcl as if you got a spectcl-5.13-010-dist tarball and had unwrapped it into the 'main' directory that is now your working directory.
These must either be installed via your system's package manager or built and installed somewhere on your system. For each dependency you must build I'll describe where the default configuration of SpecTcl expects it. By default I mean that if you install there you don't need to specify any special options to the configuration script.
- A version of NSCLDAQ is required to connect SpecTcl to the online system.
There is no default use the configure
--with-nscldaq-rootoption to specify the location of the version of NSCLDAQ you with to use. Note that this is only used to locate the helper that SpecTcl uses to get online data. - OpenMotif or lesstif. This is an X11 toolk it used by Xamine the internally
integrated displayer. Note that we recommend CutiePie now but Xamine is still
built and provided.
--with-motif-header-dirand--with-motif-libcan be used to describe where the libXm .h file are and th efull path to libXm. - libgd is used by Xamine to perfrom hardcopy
--with-gd-headerdirand--with-gd-libdircan be used to specify where this is installed if you choose to build it. - We also realy on imagemaagick for image conversions the
imagemagickpackage on Debian is suittable for this. - gri is used to produce publication quality plots. If it is not present, the Xamine just
won't print. If you build and install it yourself, specify the path to
the executable to configure using
--with-gri-dir - SpecTcl depends on the CERN Root product at http://root.cern.ch It must be built by you (or
someone). Specify the top level of the installation directory tree with
--with-rootsys - Some of the SpecTcl utilities interact with the SpecTcl REST server plugin. They use
restclient-cpp at https://github.com/mrtazz/restclient-cpp which you must build. Specify where you
installed it via
--with-restclient-cpp - We send some REST responses zipped so we need
zlib1g-devto do this compression. - The REST rersponses are parsed using jsoncpp which, in general you can install with your
system's package manager. For Debian, e.g. you'll want
libjsoncpp1for the libraries andlibjsoncpp-devfor the headers. - Xamine configuration file processors rely on flex and bison which, again, can ge installed with your package manager.
- Tcl/Tck versions 8.x are needed. On Debian we use.
libtcl8.6-devlibtk8.6-devto get everything pulled in via apt. - Several modules from the Tcl standard library tcllib are used and can be pulled in with your
package manager. On debian this is the
tcllibpackage. - Some of the standard UI scripts make use of itcl/itk and iwidgets on Debian these packages are
itcl3,itk3andiwidgets4 - We use sqlite for some configuration things so you'll want
libsqlite3-0libsqlite3-devandlibsqlite3-tclon Debian. - Documentation is generated with Docbook on Debian you'll want
xmltoon Debian. - SpecTcl includes a module that allows it to be programmed in Python. For Debian you need
python3andpython3-dev - SpecTcl is written in C++ so either g++ or an llvm based C++ compiler are needed.
If you have all of the external dependencies installed in their normal places you should be able to:
./configure --prefix=/where/to/install --with-rootsys=/where/cernroot/is/installed
This will trundle along (out of tree builds are also supported).
Once configure worked:
make
make all -jn #n like the number of processors you system has.
make install
VERBOSE=1 make check # If you want to run tests.
and you are done.