Test => Add software package dependencies for BrainScaleS(-2)… [WIP]

This initial commit simply tries to build most of the BrainScaleS-2 software dependencies. For the full "thing" we would need:

• spack download cache for source packages (as download sources sometimes change),
• maybe a spack build cache for faster builds,
• some of our environment creation code (→ via spack views),
• probably remove the duplication and directly use our singularity-based flow upstream (requires changes on our side) .

base/Dockerfile

@@ -7,7 +7,10 @@ ENV GIT_COMMITTER_NAME=platform@humanbrainproject.eu \
     MPLBACKEND=Agg \
     SDAHOME=/opt/app-root/lib/sda_flex \
     MPICC=/usr/lib64/mpich/bin/mpicc \
-    XDG_CACHE_HOME=/home/$NB_USER/.cache/
+    XDG_CACHE_HOME=/home/$NB_USER/.cache/ \
+    BRAINSCALES_PYTHON3_VERSION="python@3.8.2" \
+    BRAINSCALES_GCC_VERSION="gcc@10.2.0"
+# FIXME: This replicates information from the BrainScaleS container build flow.
 
 USER root
 
@@ -89,6 +92,14 @@ RUN echo 'from clb_nb_utils import oauth as clb_oauth' >> /opt/app-root/src/.ipy
 
 RUN fix-permissions /opt/app-root
 
+
+RUN git clone https://github.com/electronicvisions/spack -b visionary spack
+# FIXME: This replicates information from the BrainScaleS container build flow.
+RUN spack/bin/spack install -v --show-log-on-error ${BRAINSCALES_GCC_VERSION}
+RUN spack/bin/spack install -v --show-log-on-error visionary-dls-core ^${BRAINSCALES_PYTHON3_VERSION} %${BRAINSCALES_GCC_VERSION}
+# TODO: We probably also need a download cache… and we have to create spack
+#       views and expose for a custom BSS-2 jupyter notebook kernel.
+
 EXPOSE 8080
 
 USER 1000