Windows file flags

[freddiechopin]

Windows file flags

[FreddieChopin]

So on Windows proper search for executable files would only use *.exe extension as a test, right?

Does it seem possible to you that the script will work on Windows without a massive rewrite? I'm asking because I'm not a Windows user and cannot really tell whether or not trying to implement Windows support is a goal worth pursuing. If it is possible, then fixing such minor things is worthwhile. On the other hand - if it is not very likely the script will ever work on Windows - bothering with them is just a waste of time.

To make things clear - I'm open to fixing any issues which prevent the script from working on Windows (unless this complicates things too much for Linux, hopefully it doesn't), but I'm not able to support and actively maintain such use case. If it gets implemented, all I can do myself is react to issues which someone reports. Same thing as with recently added support for MacOS.

[Trass3r]

So on Windows proper search for executable files would only use *.exe extension as a test, right?

I used that as a hack for the second case: find ${installNative} -type f ! -name '*.*' -exec strip {} \; || true

Not sure how to deal with that problem in general. I think the reason for that rm in the first place is this:

bleeding-edge-toolchain/build-bleeding-edge-toolchain.sh

Line 934 in 7061c04

cp -R ${installNative}/lib/gcc ${installFolder}/lib/gcc

Which also copies cc1 and other stuff. Maybe it could be made more selective.

[FreddieChopin]

I used that as a hack for the second case: find ${installNative} -type f ! -name '.' -exec strip {} ; || true

Wouldn't it actually ignore all files with extension?

I think the reason for that rm in the first place is this: ... Which also copies cc1 and other stuff. Maybe it could be made more selective.

Yes, this is the reason for the rm. I guess instead of copying everything we could copy only files which are *.o and *.a, so the rm would no longer be required then.

[Trass3r]

1. Indeed. Which is fine for the Linux version.
2. something like a9d3bc9, resulting toolchain untested though