Last week work continued on the unfinished things that couldn't make it to the 1.9.9 stable release. DaniJ has been focusing on refactoring the libdeng1 file system in order to address a bug related to multithreaded access to cached data. I spent some time updating the scriptsys branch with the latest changes, and fixed issues in the failed build 646. The MP test servers, whose compilation was also failing, are now back online.
As we are now in the beginning of the development of 1.9.10, the roadmap is wide open for replanning and fine-tuning of direction.
One thing that has been bugging me and many of you for a long time now is the gameplay bug manifesting as the stuck column in Doom E1M7
. I will be disappointed if we don't have a solution for this by the end of the year when the stable 1.9.10 is scheduled for release.
One long-term goal for us is to get rid of the separate launcher app. To further this goal, the plan is to start removing functionality from Snowberry when equivalent configuration features are added to Doomsday itself. I intend to start this work by removing the display configuration options from the launcher at some point during the coming weeks.
Another long-term goal is to enforce proper client/server separation in the engine. The next step towards this (so called unified networking
) is to slice the Doomsday executable in half so that the dedicated server executable and the GUI client app are independent of each other. We already have the prerequisite library components in place (libdeng1 and libdeng2) into which shared functionality can be relocated. This is a fair bit of work, though, as it touches the entire engine code base, so it's likely I'll be working on this in smaller gradual steps.