Permalink
Fetching contributors…
Cannot retrieve contributors at this time
312 lines (218 sloc) 9.15 KB

Mastr-MS Development

This document is intended for developers who wish to change Mastr-MS or see how it works.

Mastr-MS is Django web application and works with both PostgreSQL and MySQL.

Source Code

The Mastr-MS code is hosted at BitBucket.

https://bitbucket.org/ccgmurdoch/mastr-ms

Git is required to check out the code.

Issue Tracking

All new bugs/feature requests should be submitted to the issue tracker:

https://bitbucket.org/ccgmurdoch/mastr-ms/issues/new

If all significant changes have a ticket associated, then release notes can be accurately generated.

Mastr-MS was originally hosted on Google Code, where there still remains a small bug list:

http://code.google.com/p/mastr-ms/issues/list

Development Environment

We develop on Ubuntu. Installing the following packages is recommended:

sudo apt-get install postgresql-server \
    ipython python-pip python-sphinx \
    python-wxgtk2.8 xvfb xserver-xephyr

Packages helpful but not really required:

python-django \
python-werkzeug python-django-extensions \
python-selenium chromium-chromedriver \

Building a CentOS RPM

The RPM build is unlikely to work unless done under CentOS. Assuming the Mastr-MS source code is checked out at /usr/local/src, you can build an RPM in more-or-less the normal way by running these commands:

CCGSOURCEDIR=/usr/local/src
export CCGSOURCEDIR
cd $CCGSOURCEDIR && rpmbuild -bb centos/mastrms.spec

The spec file requires CCGSOURCEDIR to be set. It will download all the python dependencies with pip, create the RPM, and output it to ~/rpmbuild/RPMS (or the location you have configured in ~/.rpmrc).

Building a Debian/Ubuntu Package

To get started, you need the build dependencies:

sudo apt-get install dpkg-dev
sudo mk-build-deps -i debian/control

To rebuild the debian package, run:

dpkg-buildpackage -us -uc

The packaging method is standard Debian, except for the use of dh-virtualenv.

Uploading to the CCG Apt Repo

To upload, you need the following in ~/.dput.cf:

[DEFAULT]
default_host_main = ccg

[ccg]
fqdn = staging.ccgapps.com.au
method = scp
login = ubuntu
incoming = /data/aptrepo/repo/incoming
post_upload_command = ssh ubuntu@staging.ccgapps.com.au 'aptrepo update'

Then run the dput command to upload:

dput ccg ../mastr-ms_1.13.0_amd64.deb

Make sure you remember to include ccg in the command line or it will FTP upload your package into Debian incoming.

Upgrades and database migrations

The dbconfig-common system will automatically migrate the database for the sysadmin. To let that happen, you need to define which Debian package version the migrations correspond to.

The mapping of versions is in the script debian/bin/mastr-ms-migrate.

How to build the sync client

The support libraries and binaries are in the mdatasync_client/client/supportwin32 directory.

You should be able to build the client on a 32-bit Windows XP box, by installing these resources as described below.

Initial Setup of build environment

  1. Copy client and supportwin32 over to the windows machine.
  2. Install Python27 (utf8 encoding for json is broken in earlier versions)
  3. Install wx for python 2.7
  4. Install py2exe for python 2.7
  5. Install NSIS
  6. Install 7Zip
  7. Extract the EnvVarUpdate extension
  8. Copy the extension into c:\program files\nsis\include
  9. Open a shell
  10. Change dir to the client dir

Building the code

First, make sure you have updated the version number in version.py to be unique, and sequentially higher than previous ones.

On Windows box, get the latest client dir. Run:

c:\Python27\python.exe setup.py bdist_esky

Then unzip the build you just did (in dist/) so that the files are available to the installer.

If you then ran the nsi file (right click on the file and choose 'Compile with NSIS'), it would make an installer, using the build you just did (in /dist) as a source. So that is how you make an installer.

If you want to publish just the update, you take the .zip that is generated in the dist directory, and scp it to the distribution URL. Currently, this is on S3, under http://repo.ccgapps.com.au/ma/

As long as the version.py version was incremented, a new version will be available to esky.

Errors

When building with Esky, if you get the message cannot access ../main.py. The file is in use by another process/, this probably means your python code fails a syntax check or has some other runtime error.

Running Tests

Testing requirements:
  • wxpython
  • Xvfb
  • selenium
  • chrome/firefox webdriver
  • splinter
  • dingus
  • python xvfbwrapper

Command to run:

./manage.py test --exclude=yaphc --exclude=esky --exclude=httplib2

Testing

Mastr-MS contains some system tests as well as unit tests. The current test classes are:

The WxPython client is tested using the :class:`TestClient` class. :class:`TestClient` runs the client in a thread and allows "clicking" on GUI buttons by calling the associated event handlers.

.. autoclass:: mastrms.mdatasync_client.client.test.TestClient
   :members:

To generate sample data files like the lab instrument software would create, use the :class:`Simulator` class. Simulator can be run as a standalone WxPython GUI, or operated programmatically through :class:`Worklist`.

.. autoclass:: mastrms.mdatasync_client.client.Simulator.Simulator
   :members:

.. autoclass:: mastrms.mdatasync_client.client.Simulator.WorkList

For testing you usually don't want to do actual file transfers, but you do want to check that rsync was called. For this, we put a mock command into the PATH which can be queried from the test cases.

.. autoclass:: mastrms.mdatasync_client.client.test.fake_rsync.FakeRsync

Documentation

The documentation is in :ref:`Sphinx <sphinx:contents>` format under the docs subdirectory of the source. To build it, simply run:

make html

Product Overview

This list, written by Brad, may be useful in understanding the system.

Goals of MASTR

  • To provide a web based tool for experimental design, sample metadata configuration, and sample data acquisition.
  • To enable researchers and scientists from geographically separate institutions to work together on experiments, analysis, and to be able to share results and outcomes.
  • To enable institutions to provide quotes for analysis work to third-parties, with automatic linkage through to the relevant projects and experiments.

Features of MASTR

  • User / Group administration
  • Experimental design, catering for:
    • User roles and access control
    • Sample origin metadata
    • Sample timeline and treatment metadata
    • Sample tracking
    • Sample information import / export via CSV
    • Standard Operating Procedure upload
    • Run creation, generating worklists for the purposes of instrument automation.
    • Fully customisable rules system for worklist generation
    • Sample blocks, order, randomisation, and solvents/blanks can be specified as a programmable template.
    • Worklist rulesets can be rolled out per individual, or shared with groups or the entire institution
    • Rulesets can be branched and cloned
    • Runs and Experiments can also be cloned for convenience.
  • Data acquisition, consisting of:
    • A program which runs on the computer connected to the instrument which processes MASTR worklists
    • The program will check periodically for filesets related to experiment runs being performed for MASTR
    • The sample data is compressed and uploaded to the MASTR-connected storage, and optionally archived on the client machine.
    • The sample data is then securely available to relevant users through the MASTR web interface for viewing or download.
    • Full end-to-end data acquisition, from experimental design to sample file access.
  • Quote requests and tracking
    • A system for third parties to request quotes for analysis work of any of the institutions in MASTR
    • Institution Administrators or Node Representatives can review the requests and service replies, with optional PDF attachments.
    • Full quote event history is maintained.
    • Quotes in the system can be linked to Projects / Experiments
  • Modern technologies
    • Able to be accessed in all major web browsers
    • Lightweight and powerful UI
    • Open data formats and transports used (rsync, json)
    • Open Source code repository

Auto-generated Documentation

.. toctree::
    tests