Close observers of the Xtext project will have noticed that the number of contributions like bug fixes or new features has continuously dropped over the past months and even years. The same is true for the number of people contributing. In general this is no issue as Xtext is a mature and a stable framework. On the other hand we were not able to get rid of all technical debt and bring the more future-proof part of Xtext like its LSP integration and the native Web support to the same quality level as the Eclipse part.
What kept the team busy was basically the bare minimum that needs to be done to keep up with version changes in the outside world and the necessary steps to be part of the Eclipse simultaneous release. This basically forced us to build a release every three months (adding up to 16 “releases” a year together with all the M1, M2 and M3 milestones) and to adapt to continuous changes in the Eclipse build infrastructure.
This produces a steady stream of tedious, boring and to a big extent manual tasks combined with time pressure whilst producing zero benefit beyond being part of simrel. Note that all this lasts on the shoulders of very few people. At the same time most of our customer projects don’t follow the Eclipse simrel release cycle and update every three months but rather do updates of Xtext and the Eclipse versions just from time to time. This implies even less reason for the work being done.
Over the last year we tried to advertise for either more people to contribute to the maintenance or to buy maintenance/support contracts from itemis to fund the work: Unfortunately, with very limited success.
In conclusion, being part of simrel does no longer make sense for us. We will stop the effort. Eclipse 2021-03 is the last train that leaves with Xtext onboard. For the future we will drive the development of Xtext and do releases only when an internal product or customer project requires it or a 3rd party hires us to do so.
Comments