Development update — robustness and performance work in progress

[TMailletFR]

I’ve spent most of the past week focused on robustness and performance rather than adding new features.

The current development cycle is centered on stress-testing ProjectEngine on much larger schedules and fixing the architectural issues that only become visible once the workload grows significantly. The main test case has gone from around 60 tasks to roughly 1,200 tasks, with thousands of rendered objects and more than a thousand dependency links.

A substantial amount of work has already gone into:

improving Gantt rendering and dependency geometry;
reducing unnecessary Excel/COM interaction;
strengthening incremental and local updates;
improving cold-start and update lifecycle behavior;
simplifying ownership inside the rendering pipeline to avoid duplicated work;
hardening scheduling and analytics behavior under larger dependency networks;
validating performance and correctness across both normal-sized and stress-test schedules.

Some of this work has uncovered architectural weaknesses that were not visible on smaller projects. Because of that, I’m deliberately holding the next release until the current stabilization and performance pass is complete rather than publishing an intermediate build.

A new release will be published as soon as this work is cleared.

The goal is not only to make the 1,200-task stress test faster, but to make sure the same changes keep normal-sized projects fast, predictable and robust.

The current stable release remains the recommended version until then.

Thanks to everyone who has been testing ProjectEngine or sharing technical feedback. Several external discussions and suggestions have directly influenced the current optimization work.