Markdown renderer incorrectly handles single line breaks

[the10thWiz]

Under typical markdown, single line breaks should be treated as whitespace (and combined with adjacent whitespace into a single space). However, in many places, RedReader simply eliminates single line breaks, leaving no spaces between words.

I noticed this issue on this post: https://reddit.com/r/talesfromtechsupport/comments/1uo1f3f/the_reports_are_all_wiggly/, which does render correctly in both the old and new Reddit websites.

From what I can tell, this only effects some posts (although this may be related to how the post was created, since Reddit's fancy editor might generate extra characters that allow RedReader to handle the markdown correctly). I haven't seen an example of this issue in comments, so they may be unaffected.

[axocomm]

Kind of seems like it might be how it was created, as it looks like the original selftext as well as selftext_html do include some \ns where they seem to be out of place (indeed I imagine a by-product of how the text was pasted into the "fancy" editor). If you hook a simulator up to HTTP Toolkit you can see some goodies, e.g.

• raw selftext from the API response (/comments/1uo1f3f/.json?sort=confidence):

• selftext_html:

In both cases note the single \ns, e.g. consultant at the\nComputer Center and never used a\ncomputer

I'm not yet 100% sure whether RedReader is simply rendering the HTML in the latter or doing its own Markdown processing (I assume the former should the HTML be present in the Reddit API response) but maybe there needs to be some adjustment to e.g. the HtmlReader such that it doesn't just strip out those \ns; in an ideal world maybe these would be handled by the server-side Markdown renderer before producing that selftext_html but this is modern Reddit we're talking about 😛

(E: of course expected single line break behaviour is mentioned in the OP but wondering if Markdown itself on Reddit's end should be the one to mitigate that -- if so it's not, at least in the selftext_html for this example post (i.e. single \ns still present in the selftext_html)). Running the selftext through Pandoc does seem to strip those characters out, e.g. a pandoc -o - /tmp/foo.md on that Markdown:

<h1 id="prologue">Prologue</h1>
<p>When I was in college I worked as a helpdesk consultant at the Computer Center. This was so long ago that it was routine for a student to come into the office and announce, “I have never used a computer. I wouldn’t even know how to turn one on. But my professor says we have to do our papers with <em>word processing</em>. Please help me get started.”</p>
<!-- ... -->