Update 2017-03-30: Remark on NewsBlur
Update 2017-04-23: Added link to blog article with my decision and changed I made.
In the past, Christoph did inform me of many cases where the Atom-feed of my blog had formal issues so that his reader did complain.
I learned that the XML of a valid Atom feed is quite different to the HTML code for the browsers.
By the way: The feed aggregator of my choice and my feed app which is connecting to the aggregator did not have any issue at all whereas other clients stopped working.
Case by case I fixed XML issues, mostly via search-and-replace of things like
mdash to its numeric variant or fixing HTML-snippets of Twitter or Amazon that were valid HTML but not valid feed-XML.
A way to fix this would be to use a library which converts my HTML to valid XML for the feeds. A short research did not result in a satisfying library for the job. Most libraries do much more, resulting in many additional dependencies that I tend to keep at a minimum for lazyblorg.
Therefore, I am thinking of switching to links-only-feeds. Such feeds would contain everything as now except for the content of the article body. This would eliminate all feed issues for once. A downside would be that you can not read these lines (the content) within your feed aggregator. Instead, you'd have to follow the included URL and open the articles in your browser or configure your feed reader to show the underlying web page.
What do you think about it? Does it break any of your workflows? You don't care? Drop me a line so that I can think about alternatives.
Update 2017-04-23: Read this article on what I decided to change for my feeds.