-
Node ⇐16.17.1
-
Either a global installation of
yarn >=1.17.0
, orvolta >=1.1.0
-
With such a version of
yarn
globally installed, allyarn
commands executed in this repo will be delegated to the "pinned"[email protected]
installation included in.yarn/releases
. -
With such a version of volta installed, the required version of
yarn
will automatically be installed.
-
It is recommended to install volta, as it will keep your local toolchain in sync with the repository.
This repository is a "monorepo", using the nx build system and yarn@3 dependency manager. The source code is divided into one of three types of packages: "config", "lib", and "site".
-
"config" packages (
./packages/configs
) include various build and development configurations and tools. -
"lib" packages (
./packages/libs
) include standalone library and webapp source code. -
"site" packages (
./packages/sites
) include complete website source code.
Many "lib" and "site" packages include development tools, such as development servers, test scripts, etc.
Many commands require a reference to the package name. All package names are currently prefixed with @veupathdb/
. For example, the eda package name is @veupathdb/eda
.
The repository is currently configured as a "package based repository". This is subject to change, in the future (see https://nx.dev/concepts/integrated-vs-package-based to read about the differences between package based and integrated repos).
The following tasks are common performed by developers. This serves as a reference guide, and is not exhaustive in any way. If you feel something is missing, create an issue, or open a pull request.
All commands are expected to be run in the repository’s root directory, unless otherwise specified.
"Site" projects are stored in the packages/sites
directory. Each one corresponds to a "cohort". For example, packages/sites/genomics-site
contains the code used to build the client code for a genomics website.
Each cohort contains a .env.sample
file. Copy this to a sibling .env
file. You may need to modify some values, such as usernames,
passwords, etc. You can also specify a website to use for various services. Typically a deployed QA site will suffice, but you can also run a "local backend" via various methods. (TODO: link to relevants docs/repos).
Once you have created a .env
file, you can run the local dev site with the command:
yarn nx start <package name>
For example, if you want to run a local clinepi site, you would run the command:
yarn nx start @veupathdb/clinepi-site
Once the website has been compiled, the dev server will output some build statistics and automatically open a browser tab. You can kill the local dev server with CTRL-C
in the terminal where you started it.
When a local dev site is running, it will detect when build dependencies are updated and reload the active webpage.
Note: the following refers to code tracked by the monorepo, and not third-party dependencies from npm.
There are two types of souce code that can be updated:
-
Source code within the package being served.
-
Source code within a dependent package.
Changes to code within the package being served will be detected automatically. The local dev service should reompile the affected module and reload the website without intervention.
Changes to code within a dependent workspace package will require a build command for the local dev server to detect the change.
For example, if you change code in packages/libs/eda
, you will need to run this command in a new terminal window, from the repository root:
yarn workspace @veupathdb/eda build-npm-modules
Once this command completed, the running dev server will see the updated build artifacts, recompile the website, and reload the webpage. There are some cases where this might not work as expected, such as if the recompile step fails. In those cases, you may need to restart the local dev server. You can monitor the progress of the recompilation step in the terminal where you started the dev server.
From the root package.json
:
yarn nx run <workspaceName>:<scriptName>
For example, you can start the MultiBLAST dev server by running
yarn nx run @veupathdb/multi-blast:start
When running a development server (such as yarn nx start @veupathdb/eda
or yarn nx start @veupathdb/clinepi-site
),
use the following command to rebuild changes made to dependencies, and to have the dev site reload with the changes:
cd packages/libs/<package>
yarn build-npm-modules
Note: You may need to manually reload your website to see the changes the first time.
Using the equivalent nx
command (yarn nx build-npm-modules @veupathdb/<package>
) has proven inadequate in this scenario.
Directory: packages/libs/eda
You will need to configure the server with a packages/libs/eda/.env.local
file that sets various environment variables.
For more documentation see the package README and this sample file.
Directory: packages/sites/{site name}-site
Copy the packages/sites/{site name}-site/.env.sample
file to packages/sites/{site name}-site/.env
and configure the new file with passwords and the desired backend for the site.
Run yarn
to update dependencies if necessary.
Run the command yarn nx start @veupathdb/{site name}-site
. For example, to run the ortho site use yarn nx start @veupathdb/ortho-site
.
The Client Bundle Server is a Docker image based on NGINX that is used to serve VEuPathDB client code over HTTP.
As the client code comes in 2 flavors (bundles), legacy and modern, this NGINX server has an internal path rewrite based on the requesting browser’s user agent string to the appropriate client bundle component on request.
This means using a modern browser, requesting the file
genomics/site-client.bundle.js
will cause the server to actually return
modern/genomics/site-client.bundle.js
whereas requesting that same file from
an older or unsupported browser (such as CURL or Postman) the server will return
legacy/genomics/site-client.bundle.js
.
Whether a browser is considered modern or legacy is dependent on the version of the browser compared to a RegEx constructed by the browserslist-useragent-regexp library using the input query constructed in the browserslist-config package of this repo. (See index.js for the raw queries)
The docker image is based on NGINX-Perl and includes NodeJS for executing a script based on browserslist-useragent-regexp that determines which path a specified file should be served from.
The image build is multi-staged with the first stage compiling primary contents of this repository, and the second stage setting up NGINX and the secondary JS script included in the docker directory (makeSupportedBrowsersScript.js).
Content is served from the following paths from the root path used to reach a running instance of the built Docker image:
{URL}/clinepi/{target-file}
{URL}/genomics/{target-file}
{URL}/mbio/{target-file}
{URL}/ortho/{target-file}
These paths correspond to the following container internal paths:
/var/www/legacy/clinepi/{target-file}
/var/www/modern/clinepi/{target-file}
/var/www/legacy/genomics/{target-file}
/var/www/modern/genomics/{target-file}
/var/www/legacy/mbio/{target-file}
/var/www/modern/mbio/{target-file}
/var/www/legacy/ortho/{target-file}
/var/www/modern/ortho/{target-file}
The Docker image may be tested locally by performing the following steps from the docker/ subdirectory:
-
Build and Start the image:
make docker-build make docker-run
-
Using your favorite HTTP request making tool such as Postman, CURL, or a web browser, make a request to http://localhost/genomics/site-client.bundle.js.LICENSE.txt . If the service is working you should receive a LICENSE text file’s contents as the response with a 200 status code. If it is not working you will receive a 403 or 404 error.