Posted on 2016-12-03 Edit on GitHub
I traveled with Funding Circle to Clojure/conj 2016 in Austin, Texas this past week. My first Clojure conference had been Clojure West in Seattle, Washington back in in April 2016, to which I had gone alone still working for Epic.
Traveling as a group to a conference is much more interesting–you'll talk to more people, especially those that are interested in your company. As Funding Circle is one of the few companies committed to using Clojure for more than small one-off projects, interest in the company was high.
There were a few interesting talks this year, though not as many that I found compelling as at Clojure West. Many talks were oddly specific; instead of providing high-level overviews that leave the audience excited to try new things when they get home, they were post-mortems of projects done at particular companies for narrow purposes. This sort of talk is great for a local meetup, but not the major conference of the year.
Rich Hickey's keynote, as expected, did not disappoint. He dissects the problems with Semantic Versioning (semver), one of the most popular versioning schemes, and really nails it on the head with the observation that versioning semver itself destroys the semantics of anything that uses it.
The key take-away is that projects, from open-source libraries to commercial APIs, should never break backward compatibility. This is a core value of Java and Java-related technologies, so it's not surprising that Clojure is advocating this as well.
What Rich envisions goes further than this, however; he wants to use
clojure.spec to make guarantees about compatibility of functions, namespaces, etc., which would require packages and artifacts to be build differently. I'm not sure how practical this is–one of Clojure's core values is interop with the host platform/ecosystem, and incompatibilities with "the Java way" will not prove popular.
Overall, this was a fun trip, and I look forward to bigger developments at future Clojure conferences.