Re: "*** INTERNAL: readdir: Invalid argument" error

[Eli Zaretskii]

> Date: Wed, 15 May 2013 11:35:47 -0700
> From: David Boyce <address@hidden>
> Cc: bug-make <address@hidden>
> 
> 2. What to do about the more general case of an opendir/readdir
> failure, when making legitimate use of the caching function, is still
> undetermined. Is it better to fail, as now, but with a somewhat better
> error message or continue with a warning that files in that directory
> will be ignored?

I still think a fatal error is too drastic here.  A warning, or even
just a debug message (enabled under --debug) should be enough.  After
all, chances that whatever is causing the problem will actually
interfere with the build are nowhere near 100%.