Re: bison segv under Cygwin 64 at fatal-signal.c:318

On 2021-09-14 10:09, Akim Demaille wrote:
> Sorry for the slow response...  I'm still hoping we could reproduce
> this failure outside of Bison, so it took me quite some time to wrap a
> tarball of gnulib with its test suite.  My first attempt ran almost all
> day long (C-c to rescue), and generated something like a 1GB
> directory...  My second attempt, with
> ./gnulib-tool --create-testdir --dir=./gnulib-with-tests
--single-configure --with-c++-tests --without-privileged-tests
> was much faster: "just" about 2hrs :)
> then, ./configure && make dist, and voilà, the following tarball:
> https://www.lrde.epita.fr/~akim/private/bison/dummy-0.tar.gz
> Could you please try it?  With some luck, some tests will fail,
> which we'll be able to report to gnulib.

Well I used my normal Cygwin packaging environment and process with 
--debug flag to see some of the internals, and 43 tests failed, but I 
can not see enough of the test details to figure out why l/dtoastr would 
fail with permissions problems, and why other tests show the symptoms 
they do?

The process took my system ~7.5 hours, and I attached all the relevant 
logs I can find (sanitized for my protection, as I noticed some leakage 
in previous logs).

Is there some other process I should perhaps use to build and test gnulib?

Santé!

-- 
Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada

This email may be disturbing to some readers as it contains
too much technical detail. Reader discretion is advised.
[Data in binary units and prefixes, physical quantities in SI.]