CommaFeed is now powered by Quarkus instead of Dropwizard. Read the rationale behind this change in the announcement.
The gist of it is that CommaFeed can now be compiled to a native binary, resulting in blazing fast startup times (around 0.3s) and very low memory footprint (< 50M).
- CommaFeed now has a different package for each supported database.
- If you are deploying CommaFeed with a precompiled package, please read this section of the README.
- If you are building CommaFeed from sources, please read this section of the README.
- If you are using the Docker image, please read the instructions on the Docker Hub page.
- Due to the switch to Quarkus, the way CommaFeed is configured is very different (the
config.ymlfile is gone). Please read this section of the README.
Note that a lot of configuration elements have been removed or renamed and are now nested/grouped by feature. - Added a setting to prevent parsing large feeds to avoid out of memory errors. The default is 5MB.
- Use a different icon for filtering unread entries and marking an entry as read (#1506)
- Added various HTML attributes to ease custom JS/CSS customization (#1507)
- The Redis cache has been removed. There have been multiple enhancements to the feed refresh engine and it is no longer needed, even for instances with a large number of feeds.
- The H2 migration tool that automatically upgrades H2 databases from format 2 to 3 has been removed. If you're using the H2 embedded database, please upgrade to at least version 4.3.0 before upgrading to CommaFeed 5.0.0.