Instructions for setting up the CSAMA local mirror and creating the installation scripts to be sent to the participants.
** We'll worry about this later **
Run wget_repositories.sh. This bulk downloads the whole of Bioconductor and CRAN and create the directory structure required for our own mirrors of both. We only want to run this once as it's approximately 100GB of downloads, afterwards we can use rsync_repositories.sh which will only update the bits that have changed.
We maintain a list of the required packages in common_files/packages.txt. I've been doing this manually by looking at the Rmd files in https://github.com/Bioconductor/CSAMA/tree/2017 and copy/pasting. It doesn't matter if there are duplicates, so put in anything that looks like it might be needed.
If we have any additional packages that have only been created for the course e.g. data packages, these will need to be hosted in our own repository. A plain source package folder (not built or zipped) should be placed in common_files/additional_packages/. A source tarball will be built from this and put in the correct place under csama_repo. Remember you'll need all the packages listed in the DESCRIPTION file (Depends, Suggests, Imports, Extends) installed on your system in order to build it successfully.
If the package has C code or similar that needs to be compiled, we'll have to make Windows and Mac binaries. Hopefully this doesn't happen, but see if the package author can provide them. If not someone in the lab must have a Windows machine we can borrow. These should be copied manually into common_files/csama_repo/bin/macosx/contrib/3.4/ and common_files/csama_repo/bin/windows/contrib/3.4.
So far only the devel version xcms needs to be built like this, and it's already done, so hopefully nothing more is needed for this section.
We can then go into the user_installation folder and run create_install_script.R.
cd user_installation
./create_install_script.sh
This first called Rscript on the file common_files/calculateDepedencies.R. This processes the list of packages, builds the source versions of any additional packages, and creates install_packages.R which can be provided to the students as an install everything script.
The shell script then tries to scp the installation script to a location on the Huber group website. Currently this is at /g/huber/www-huber/users/msmith/csama2017/ and contains both the installation script and the copy of repository containing our additional packages.
There is a draft of the email to send the participants in user_installation/InstructionsEmail.txt. Feel free to use any or all of it.