Re: testing Context O/J

[Simon Josefsson]

"Abdulrahman I. ALGhadir" <address@hidden> writes:

> Dear Simon,
> I am Abdulrahman from SaudiNIC and I am willing to help in testing the
> part which is related to Context O/J for R2L languages such as Arabic.
> I am going to inform you with details but do you have certain
> scenarios in your mind for test or just do my best?

Hi Abdulrahman, and welcome!  Your help is really important, I don't
have your expertise.

What would help if you have a bunch (say 10-20) common R2L domains to
make sure we don't introduce any serious regressions.  You want to add
them to tests/test-lookup.c like the following:

static const struct idna idna[] = {
...
  {"\xe6\xb5\x8b\xe8\xaf\x95", "xn--0zwm56d" },

The first string is escaped UTF-8 and the second is the ACE encoding.

Second, and perhaps more challenging, is to prepare real corner-case
test vectors that triggers the IDNA2008 rules related to CONTEXT O/J and
BIDI rules.  What we would want is to make sure all paths in the code is
tested.

I'm going to prepare code coverage charts for libidn2, which would help
to identify which parts needs testing -- that will help us find the
necessary test vectors.

/Simon