Stephen John Smoogen wrote:
These systems are not small tasks to keep up and running.. the
upstream code is not 'static' or backportable so you are constantly
updating to upstream to keep up with CVE security. There are also
regular schema changes and a ton of packaging items which need someone
who is going to become a discourse expert to run. Our experience has
been that we either end up having a system which is broken a lot
because it isn't being maintained or is taking up so much time that
Fedora people complain we aren't working on getting a compose out the
door.
So instead we decided to invest money into a company which pays the
authors of discourse (and we previously paid the person who wrote
askbot). That means we do lose absolute control but we do fund the
upstream.
Yet, this means that we are experiencing exactly the kind of lock-in that we
claim to free us and our users from.
Having to pay extra for some features (as the hosted version of Discourse
does it) is lock-in. (Interestingly, unlike, e.g., Gitlab, Discourse
apparently does not follow this same crippleware strategy for the self-
hosted version, they state that all features are provided as Free (Libre)
Software and at no cost (if you self-host the application). Only the hosted
version of Discourse does market segmentation.)
Having no way to migrate the data when switching to a different platform is
lock-in, too.
Kevin Kofler