Mats Wichmann

Applies to C23 as well. For grins, here's the example from the C standard: ```c #ifdef __STDC__ #define TITLE "ISO C Compilation" #elifndef __cplusplus #define TITLE "Non-ISO C Compilation" #else...

Tagged this with variantdir because that (in the old form BuildDir) appears in the SConscript.

Something about this sounds familiar - wasn't there some ordering issue? #2877 has one of these.

> The original filer is using `glob.glob('*.cpp')` which doesn't guarantee order staying the same between runs... That could have something to do with this.. I guess we've each "ported" this...

> Bear in mind that I don't know what I'm talking about. > > The runs that pass seem to consider node `build\\debug\\lib\\core\\static\\core` before any individual file nodes (e.g., `build\\debug\\lib\\core\\static\\corF.cpp`)....

We always say "the taskmaster will figure it out", but there have been more than one case where things that are essentially "side effects" which also become sources don't work...

You're absolutely right, that is the difference, it fooled my old tired eyes, because after I moved the `/c ...` stanza in one copy, they were the same length and...

> This also has the problem where the VariantDir()'s source is a parent of the variantdir.. > > `staticlibenv.VariantDir(builddir+"/static", ".", duplicate=0)` @bdbaddog you keep saying this, but there are so...

I've wasted a little more time poking at this, and I'm pretty convinced everything is consistent coming out of the sconscript phase - the PCH builder has been invoked the...

> @mwichmann When you have time, would you please review to see if the following makes any sense. > The difference between the builds that fail and the builds that...