So I investigated the historic testbots results on master gcc8.4.0
And those looked innocent, good so far.
So I made a big-plan to port back from 0.4.15-dev-2951-g5d8e834 whatever would apply cleanly on top of 0.4.14-RC-105-g6375090.
TheBigPlan_unifiedDiff.patch (which is only a subset of Timos commit, because some of the 'broken' implementations were not even existing in that older branch).
And then I checked the resulting buildlogs before and after 'theBigPlan' on MSVC2010SP1 16.0.40219.1, targetting x86:
The result was very disappointing, because I realized that Timo Kreuzer was mainly targetting to fix x64 build-issues with that commit, and the only win for my older branches are the fixes for
A pity, not much benefit for the old branches then.
But since I did all this effort, I decided to backport that minor fraction of improvements at least: crt_TheSmallPlan.patch