David Allsopp

I would certainly read `HOME` first (i.e. read something from the environment) since, as with Unix, it allows you to temporarily tell one program to use a different directory. Regardless...

The source of _this_ problem is `print_string` - all the calls in main.ml use binary `Unix.file_descr` so test.ml should only receive `\n` on stdin, therefore the fact that stdin is...

A couple of quick thoughts triggered by another discussion. I agree that hidden (though of course documented!) command line switch seems the best bet, but how about making it one...

In terms of the output formats, it might be useful to have something which is "trivially" parsed in a shell-script - just thinking for example of [src/state/complete.sh](https://github.com/ocaml/opam/blob/master/src/state/shellscripts/complete.sh) in opam, which...

Two brief comments: - One solution to the lack of `extra-sources` filtering is to put the URL in another package and depend on it (see the way the `ocaml-config` package...

I just discovered I was being bitten by this too - making `HAVE_INT128` is definitely not helping with gcc 9.2.0 on Cygwin 3.1.7 in a Docker container on Windows 10...

This and forcing `-O0` is working fine, for an unoptimized value of "fine"!

I haven't tested if `-O0` on its own is sufficient, no (it looks like it probably will be). I was going on the assumption that intrinsic support for `__int128_t` is...

I should have had a more in-depth glance, `__SIZEOF_INT128__`, is provided.

OK, I've dug a bit further. The problem is alignment, and it's not Windows-specific, I expect it's just not been seen on Linux or elsewhere yet. Here's the relevant assembly...