Nicklas Larsson

This ends up with mixed `/` and `\` which I believe was part of the original problem: `ERROR: Unable to make mapset element .tmp\unknown (C:\Users\runneradmin\nc_spm_full_v2alpha2\__general_g_remove_test_g_remove_fv_az587_258_3624/.tmp\unknown`. This particular forward slash originates...

Let us see the results of CI runners first. There are a number of hardcoded cases with `/` for files in $GISBASE, e.g. `etc/symbols`, `etc/proj` etc. Perhaps it is not...

> Portability is not an issue. `/` is a perfectly fine directory separator on Windows ([AltDirectorySeparatorChar](https://learn.microsoft.com/en-us/dotnet/api/system.io.path.directoryseparatorchar?view=net-8.0#examples)). That is unfortunately only valid for .NET API, perhaps https://superuser.com/a/176395 may throw some light...

> > `strcat(element, "/");` should be > > `strcat(element, HOST_DIRSEP );` > > Would it harm to change this, @hellik (others) ? Wouldn't do any harm I think, it is...

> Apparently, not the correct way. Does maybe @nilason have a hint how the pointer magic is done correctly? `HOST_DIRSEP` is defined as char, whereas `strcat()` expects a string.

That looks good to me, sounds like a plan for a new PR :-).

> About the question in a review comment if it occurs only on Apple silicon arm or any arm, does anyone have a raspberry pi available to try this on?...

> I tried adding a `-fno-fast-math` cflag and compiled `r.texture`. The Arm64 Mac gives the same results as those from x86. I believe we should put this flag in a...

Now, this is an issue again on the CI runner.

CI just bumped the clang version from 19.1.7 to 20.1.1. test_rtexture.py/test_rtexture_parallel.py: ``` mismatch values (key, reference, actual): [('max', 45368496, 45368492), ('mean', 2248724.38215788, 2248724.35922656), ('variance', 2332049495199.41, 2332049429651.92)] ```