How-to



Overview

The following will be of interest to the collaborators interested in contributing to this site or involved in its maintenance.

We use GitHub to manage the code for this site. Please take a look at the GitHub repository to get an idea of the general organization of the data, layouts and supporing logic. The idea is to shape the code and content in a way that is easy to navigate and modify. The following sections explain how this is achieved. Development of this site involves the following aspects:


The Menu System

Entries in the navigation bar on top are named similar to the folders in this project, for example the “Software” is associated with the folder “_software” etc. These folders are treated as “collections” by the Jekyll framework and they need to be declared in the main configuration file _config.yml. As one can see in the navigation bar code, its menu entries are generated automatically based on the content of the file _data/menus.yml. The content and order of top entries in the navigation bar as well as content and ordering of all items in all dropdown menus are defined in that file.

If a menu entry in _data/menus.yml is an external link, there is no additional source code (i.e. markdown page content is not necessary). Examples are easy to spot in the menu definition file _data/menus.yml.

Menu entry descriptions in menu.yml can also have optional attributes:

Front Matter

“Front Matter” is a piece of YAML code (typically short) embedded on the very top of a Markdown-formatted document which can contain any sort of data, for example variables used to render the page or define a tag so it can be more easily referenced in this way. The data in the Front Matter is accessible in a way similar to object attributes. Importantly, when present the Front Matter serves to inform Jekyll that this particular page needs to be rendered into HTML for inclusion in the site being built. Otherwise it won’t be included in the site. In summary, most of the files used to build the site are expected to be in the Markdown (MD) format, and each MD file is to be equipped with the “front matter” section, which could look like the excerpt below.

---
title: My Cool Software
name: my_cool_software
category: mysoftware
layout: default
---

Helper macros

Tables

All tables on this site have added CSS styling to enhance appearance and readability.

To define tables within pages one can use built-in Markdown table styling capabilities, for example this snippet of Markdown code

| Language                 | Lines of code                       | Description |
|--------------------------|-------------------------------------|-------------|
| C++                      | 101                                 | Module 1    |
| Python                   | 37                                  | Module 2    |
| Ruby                     | 53                                  | Parser      |
{:.table-bordered}

…will be rendered as

Language Lines of code Description
C++ 101 Module 1
Python 37 Module 2
Ruby 53 Parser


Main disadvantage of this method is difficulty in editing large table cells (i.e. cells with a large amount of content). There is also an option to embed HTML directly. Both Markdown and plain HTML options will work but have limitations for example multiple tables present on the same page may end up having diferent widths, and column within individual tables may also have inconsistent column widths. This detracts from the page readability.

For greater control and ease of editing data there is also a macro called big_table.md deployed on this site which is illustrated in the following example. Assume the following content has been added to the front matter of a page:

tables:
  table1:
    width: 20%
    columns: [30%, 70%]
    headers: [col1, col2]
    rows:
      - [ a, b ]
      - [ x, y ]

…and the source of the page contains:

{% include layouts/big_table.md table=page.tables.table1 %}

This will result in the following content on the page:

col1 col2
a b
x y


Note how the table and column widths are configured in the example above. More tables can be easily added by putting the content in blocks of the front matter and then referring to them using the macro, as needed.

Individual cells can contain long strings and/or strings split over multiple lines as long as these are double-quote delimited in the front matter. Example:

tables:
  table2:
    width: 20%
    columns: [50%, 50%]
    headers: [col1, col2]
    rows:
      - [ a,
      "this is an example of a very wide column which can potentially span many text rows in the same cell" ]
      "two lines
      in YAML" ]
col1 col2
a this is an example of a very wide column which can potentially span many text rows in the same cell
x two lines in YAML


Frequently used external links are located in the following registry: _data/links.yml.

They can be easily inserted in pages in a consistent manner by their mneumonic name using a macro as decribed below. This simplifies handling user-unfriendly links and ensures they remain the same across the site and makes it easy to point to a different resource when necessary. For example, this piece of code

{% include navigation/findlink.md name=’Jekyll’ %}

will result in the link: Jekyll. If you would like the displayed link name to be different, the optional “tag” argument for the macro will do that. Example:

{% include navigation/findlink.md name=’github_site’ tag=’repository’ %}

will result in the following link: repository. Links will automatically open in a new tab. There is a similar macro for use with links to internal pages on the site. Example:

{% include navigation/pagelink.md folder=site.about name=’howto’ tag=’“How-to” page’ %}

will result in: “How-to” page (this happens to point to the page you are reading now). Primary advantage of this macro is that it allows renaming the source file - while keeping the same “name” attribute in the Front Matter section the links will still be correct.

Images

Users and developers have complete freedom in how they incorporate images into pages on this site. In many cases handling images will be facilitated by adding an image to the following registry: _data/gallery.yml. Note that items in gallery.yml contains references to paths of images. While any image can reside in any folder it is recommended to keep the image files under asset/images in one of the following four folders:

The site folder is used for keeping elements of the site look and feel, the software one is used to keep various software-related diagrams, and the rest are self-explanatory. Once an image is added to to the appropriate folder and gallery.yml one cab use a simple macro to refer to it by its mneumonic name and automatically add it to pages on this site, as illustrated in the following example. The PNG file containing the “news banner” image is added to the site folder under assets/images and this lines are added to gallery.yaml:

- name: news_banner
  path: '/assets/images/site/EICUG-SWG-News-Banner.png'
  title: 'EICUG SWG News Banner'
  type: logo

Then, this line of code included in the page

{% include images/image.md name=’news_banner’ width=’400’ %}

…will produce

If you are adding a software-related diagram, it should go into the “software” folder and the gallery.yml should reference it accordingly. This helps to ensure consistency of image links throughout the site and remove the need to include cumbersome HTML in the page code. It also makes it possible to move the content across folders if necessary and only make changes to the path in one place.

Documents

Similar to the logic presented above, it is preferable to create references to documents (both internal to the site but also as references to cloud storage such as Zenodo) in an organized manner. Similar to examples above, there is a registry of documents on the site: _data/documents.yml. It is quite small as the time of writing but is expected to grow. Documents can be referred to their names and links autogeneraged by using a macro such as one in this example:

{% include documents/doc.md name=’Jung-MCValidation.pdf’ %}

One of the advantages of this approach is that the source of the document can be relocated within the site or even uploaded to an external resources while the links in the site code will remain valid provided corrections are made in just one place: _documents.yml.


Managing Data

Jekyll is fairly flexible when it comes to storing and manipulating structured data. The data component of the site can reside in the “front matter” section of the Markdown-formatted files or in separate YAML (or JSON, CSV etc) data sources. The front matter approach works well for small sites. For scalability, it is recommended to rely mostly on dedicated data files (i.e. files in the “_data” folder) and keep the content of the front matter sections of individual MD files to a minimum.

The Liquid template language features a variety of filters that can be applied to the data stored in YAML and other data sources as well as in the front matter blocks of pages, and decent (not perfect) support for collections, iterators and flow control constructs.

Information assets on this site (e.g. imaged, PDF files etc) are stored in the aptly named “assets” folder and its subsiduaries. This convention should be kept going forward.

Formatting

We aim to provide a uniform look and feel across the site. To that end, whenever possible the head (title) of each page is formatted by using the standard include layouts/title.md - please look a the code for examples of its use. It’s renders the page title from the front matter sections.

Headers of sections within a page are currently formatted in Markdown as “header level 4” i.e. prepended with four hash characters like in ”#### My section header”. Take a look at the header (Formatting) of this section to get an idea of how it’s rendered.

Blockquotes are used to put emphasis on quoted text. The standard way of doing it in Markdown is to prepend the lines that need this formatting with the ‘>’ character. For example:

>Blockquote!

…will produce

Blockquote!

Language-aware code highlighting capability has been integrated and can be easily used via the fenced code blocks i.e. blocks delimited by triple backticks, with an optional shortcut for the programming language name. For example,

```bash
export foo="foo"
echo $foo
ls -l .
```

will be rendered as:

export foo="foo"
echo $foo
ls -l .

and

```python
def my_function():
  print("Hello from a function")
```

will show as:

def my_function():
  print("Hello from a function")

For C++, the opening line should be triple backticks followed by “c++” similar to examples above and the code will be rendered accordingly such as:

class MyClass {       // The class
  public:             // Access specifier
    int myNum;        // Attribute (int variable)
    string myString;  // Attribute (string variable)
};

Development and Deployment

Ruby and Jekyll

To productively participate in the development of this site one needs to learn the basics of the Jekyll framework and install it on a development machine. This is relatively straighforward but may depend on the OS specifics.

Once the environment is in place, any modification can be validated immediately since the locally running development server provided by Jekyll will render the site on the local host. Basic knowledge of the Liquid template language and in particular the “filters” that are part of it is extremely helpful.

GitHub Pages

As already mentioned, we use GitHub to manage the code for this site. A fork/pull request process is optimal, but if you prefer direct push for some reason please get in touch with the group. To get access to the code, create a clone of the website repository and move to the directory created:

  git clone https://github.com/eic/eic.github.io.git
  cd eic.github.io

The repository is configured for ‘‘GitHub Pages’’. This means that once modified contents are committed to the repository and you perform git push, the GitHub pages system will build the site and make it publicly available at https://eic.github.io/. Errors in the updated code may result either in failed deployment or a broken instance of the site. For that reason it is desirable to assess the result of changes before publishing them. To achieve this, you need to install Jekyll on your local machine, as detailed below.

Jekyll

Instructions can be found on Jekyll web site but the short story is:

Once Jekyll has been started you can view the web site by connecting to localhost:4000. Changes made to files are immediately compiled and reflected on the displayed site (at the next page load). The optimal workflow is to make changes and debug entirely locally before doing a ‘git commit’ and pushing to GitHub.

Examples of Jekyll commands

# Start a new Jekyll project
## NB. this is a little tricky e.g. you would need to pick a correct
## Gemfile and perhaps use --force
## A little experimentation is in order
jekyll new myblog

# Clean stale CSS etc, also run after modifyig _config.yml.
bundle exec jekyll clean
# Serve content while filtering out some annoying warning messages
bundle exec jekyll serve 2>&1 | grep -v ': Using'

If you get this far, you are ready to start modifying the site. A word of caution - please check the output of the jekyll compiler for error messages. Normally the code compiles and you just reload the page to see the result, but if there was an error and you are not checking the compiler’s windown the behavior of the page will be puzzling.

top