Dagobert,
> Le 18 août 2021 à 16:28, Dagobert Michelsen <[email protected]> a écrit :
>
>> BTW, please also run `make check-tests`. Because you had errors in the
>> tests of the examples, the main test suite was not run.
>
> There are quite a lot of failures:
> testsuite: 4 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
> 487 48 500 failed
>
> The log is attached:
The test case 4 fails this way:
./input.at:104: "$PERL" -p -e 's{([\0\200\210\360\377])}{sprintf "\\x%02x",
ord($1)}ge' stderr
2c2
< 1 | %header "\xf0\x00\x80\x88"
---
> 1 | %header "\xf0\x80\x88"
5c5
< 2 | \x00\xff?
---
> 2 | \xff?
8c8
< 3 | "\x00"
---
> 3 | ""
So there is something fishy going on with the nul bytes. I don't know if it's
in the input generated by the test suite, the processing by Bison, or the
pricessing of the diagnostics by the test suite.
Could you please send tests/testsuite.dir/004/input.y as an attachment? The
results of
- od -x _build/g11d/tests/testsuite.dir/004/input.y
- LC_ALL=C tests/bison tests/testsuite.dir/004/input.y 2>log
could be useful too. Thanks!
