I am glad to report that we were able to execute a successful Hackathon with 11 participants and 4 ideas actively worked on:
- Multiplayer improvements (https://wiki.flightgear.org/Hackathon_P ... provements) included two investigations and fixes.
- Make sure that models appear correctly over / on the ground. A solution for AI models has been found and implemented. In the making: check that all AI models in FGData have the correct offset assigned. To be done by aircraft developers: make sure that the offsets are correct.
- Sometimes high-res cockpits are shown, when the rest of the model is displayed at low-res. A solution was found based on distance - logic based on pixel is to be investigated further.
- Performance improvements on ortho scenery (https://wiki.flightgear.org/Hackathon_P ... ho_Scenery): works now very well with WS3.0. In addition the ortho scenery now takes precedence over grass effects on airports, which makes a very smooth visual integration. It looks stunning.
- Make properties fully thread-safe (https://wiki.flightgear.org/Make_proper ... hread-safe). Over 3000 LOC were updated making all get_value functions thread-safe. There is one bug in relation to the menu and some open questions regarding minor performance problems. Getting this done further improves programme execution correctness and opens doors for better modularisation of FG, which by itself is an important feature.
- Performance testing by replay recordings (https://wiki.flightgear.org/Performance ... recordings): A successful proof of concept was implemented in Python and demonstrated. The usage is promising and opens for better and low effort regression testing of larger changes - e.g. trial and error optimisations for osm2city mesh sizes.
During a short retrospective it was discussed, whether we really should wait for one year until the next Hackathon - or increase the frequency to e.g. 3 months or even shorter. A higher frequency might also make more room to have some Hackathons focussed more on e.g. aircraft development instead of work in the FG core - and thereby attract different participants to a social get-together.
The used Mattermost infrastructure with a linked free Jitsi video conference hosting worked very well like last year and can be made available to more/different Hackathons. It might not be always necessary to have formalised ideas registered in the wiki and have a long preparation time - however most participants feel that there is value in having ideas roughly formulated in some form, such that other participants can get attracted and that the start-up pitches/discussions can be kept short enough.
If this triggers interest and you want to organise a Hackathon - or get help with organising one based on a topic - then feel free to contact me or other participants of the Hackathon.