Markdown modeling

[bqpd]

@whoburg, @mjburton11, @pgkirsch, curious for your feedback on this proposed documentation technique.

[mjburton11]

@bqpd, could you briefly explain to me what you're doing here and how default.latex and gas_hale.md interact with the model?

[bqpd]

@mjburton11, sure thing!

gas_hale.md is all the code from gas_hale_rebuild3.py plus all the documentation in documentation/. It's a markdown file that can be:

• run as a python file per https://github.com/hoburg/gpkit/pull/547
• viewed and edited as a wiki on Github without any extra work
• compiled to this nice pdf with pandoc gas_hale.md -o gas_hale.pdf --template default.latex

default.latex is a latex template with \usepackage{bigtable} so that the solution table can be compiled in the pdf.

[mjburton11]

I think this is very useful. How are you pulling the data from documentation? I'm having trouble running pandoc error: pandoc: command not found

[bqpd]

Whoops, you'll have to install pandoc first.

The idea is that instead of maintaining separate models and model documentation, they can be combined into a single file, so I copied the assumptions from documentation/ by hand.

The rest of what was in there were constraint and variables lists, which I think we can (and should) be generating automatically. Special python comments in code blocks can skip that block or replace it with generated latex in the PDF, so all that remains is writing the python to generate the right latex.

[bqpd]

I could imagine incremently replacing the code with generated latex such that the final result is a lovely codeless PDF

[bqpd]

From @whoburg:

• I think the latex template should \usepackage{fullpage} -- right now many code blocks are running off to the right. An 80 char block should look good.