How to include code snippets in blockquotes

[ocaisa]

It is very tedious to include code snippets in blockquotes but I have a workaround, see the example below

> ## Learn to call the **GPU** package from command-line
>
> In this exercise, we'll deal with a Lennard-Jones (LJ) system as described by the
> following input file. You can vary the system size and the length of the run using the
> variables `x`, `y`, `z`, and `t`. For this exercise, let us choose `x = y = z = 60`. Since
> this is a system with `fcc` lattice, the total number of atoms should be 864,000 for
> the chosen values of `x`, `y`, and `z`. Let, `t` = 500.
> We'll call the **GPU** package from the command-line in this case. Can you prepare a
> job submission file for this system such that it enables to use 2 GPUs with 24 MPI ranks.
> Make sure that the neighbor is built on the CPUs and there is a dynamic load balancing
> between the CPUs and the GPUs.
> ```
> {% capture mycode %}
{% include /snippets/ep05/in.lj %}
{% endcapture %}{{ mycode | strip | newline_to_br | replace: '<br />', '<br />> ' | strip_html | strip }}
> ```
> {: .code}
>
> > ## Solution
> > A job submission script is shown below. Note that the number of MPI ranks is fixe
> > by `#SBATCH --ntasks-per-node=24`. You are requesting for 2 GPUs by using
> > `#SBATCH --gres=gpu:2`. Rest of the input parameters can be passed to the LAMMPS
> > executable using command-line switches. The system size can be chosen using
> > `-v x 60 -v y 60 -v z 60`, length of the run can be decided by `-v t 500`, **GPU**
> > package and the number of GPUs is chosen by `-pk gpu 2`, the **GPU** package
> > related fix/pair styles can be chosen using `-sf gpu`, and other **GPU** package
> > related keywords are chosen using `neigh no newton off split -1.0`. A `no` value of
> > the `neigh` keyword ensures that the neighbor list is built in the *CPUs*. Dynamic
> > load balancing option between CPUs and GPUs is chosen using `split -1.0`.
> {: .solution}
{: .challenge}

Note that replace is key, if the code snippet is in the solution is would change to

{% endcapture %}{{ mycode | strip | newline_to_br | replace: '<br />', '<br />> > ' | strip_html | strip }}

[ocaisa]

This didn't work, I had to go another way and create our own custom filter. This is included in #23 and can be used like:

> ```
> {% capture mycode %}
{% include /snippets/ep05/in.lj %}
{% endcapture %}{{ mycode | append_to_newline: "> " }}
> ```
> {: .code}

[ocaisa]

The last thing also didn't work on the GitHub site, so I had to revert to the standard filters. I managed to get it to work with

>
> {% capture mycode %}{% include /snippets/ep05/in.lj %}{% endcapture %}
> {% assign lines_of_code = mycode | strip |newline_to_br | strip_newlines | split: "<br />" %}
> ```{% for member in lines_of_code %}
> {{ member }}{% endfor %}
> ```
> {: .source}
>

The syntax is very very picky though, it has to be exactly like that.