Hi! I found a fix for a specific startup-speed regression affecting Geany master. It may or may not have anything to do with what people experienced in #2649, but it can be reproduced as follows: give Geany a session of ~300 open C files to restore, and otherwise a default configuration, and simply start it. Geany master: ~12-24s. Geany 1.35: ~2-4s.
This is all moot if the Scintilla v5 upgrade (#2824) is imminent, as the cause is a mistake backporting some v5 code to the LongTerm3 branch. It started with https://github.com/geany/geany/commit/4bceddb449b8727e1d8b449a5d92af5fa53a24..., was caused by http://hg.code.sf.net/p/scintilla/code/rev/6c453822904a, and can be fixed with a 1-line change I posted to https://sourceforge.net/p/scintilla/bugs/2280/. Per Neil's comments there, further 3.x releases are unlikely, so Geany's options are to either adopt the fix directly or press on with the v5 upgrade.
Please let me know if a PR with the 3.x fix would be useful. (Will any releases happen before the v5 upgrade is done?) Best, Justin