← Back to team overview

libravatar-fans team mailing list archive

Re: Libravatar tests

 

Hello,

I ran the tests again and the incorrect behaviours are fixed. Sadly it introduced a slight regression for which I opened a new issue (#20).
Thanks for your responsiveness :)

Sorry but I have no idea what “swag” means in this context. Can you enlighten me ?

Regards.

On 11/16/18 10:22 AM, Oliver Falk wrote:
Hi!

@Tristan: I cannot thank you enough! You did an amazing job by testing, providing feedback and opening issue. It's amazingly awesome to have you here! Interested in some Red Hat or Fedora swag?

@All: Is there anybody who has knowledge about Django/Python testing framework? I have already quite a lot of tests, so it shouldn't be too hard to add what Tristan provided to the test suite.

The issue that Tristan opened in my GitLab instance have now been addressed and the latest code should be running already.

Kind regards,
  Oliver

On Tue, Nov 13, 2018 at 12:11 AM Tristan Le Guern <tleguern@xxxxxxxxxxx <mailto:tleguern@xxxxxxxxxxx>> wrote:

    Hello,

    My hand-rolled tests are now converted to sharness as suggested by Lars
    Kruse, which in my opinion improved them. They are available here:

    https://github.com/Aversiste/libravatar.cgi/tree/master/regress

    As described in the README curl and sharness need to be installed. The
    tests should otherwise be easy to run (tell me if they are not !).

    There are two test files of interest for us here:

    * ivatar.t ;
    * libravatar.org.t.

    I wrote ivatar.t with what I read in its source code as well as the
    knowledge of a previous test session I did some time ago, but I haven't
    tested it effectively as I can't register my email address on
    avatars.linux-kernet.at <http://avatars.linux-kernet.at> (I sent a
    separate email to Oliver Falk for this
    issue).

    The writing of libravatar.org.t gave me some troubles for a few
    reasons.
    As I said during the last meeting it is too permissive as it happily
    accepts requests for size outside the documented bounds (1-512). A size
    of 0, or even a negative one, returns a HTTP 200 code and a 1 pixel PNG
    image. Another issue comes from the systematic redirections to
    Gravatar,
    even for trivial requests such as default=404 or
    default=http://cdn.libravatar.org/nobody.png on a non existing email
    address. Gravatar caches optimized versions of the avatars it serves,
    making them hard to compare with their original source.
    Also some of the tests are marked as “test_expect_failure” as I am not
    sure what we want to consider acceptable. For my implementation, which
    is quite strict, I don't allow empty value for the “size” or “default”
    parameters nor non-integer value such as “size=onehundred”. The
    official
    documentation[1] doesn't mention what is acceptable or not so it is up
    to us to clarify this point.

    You can find the output of libravatar.org.t and libravatar.cgi.t (my
    implementation for comparison) attached.

    Regards.

    [1]: https://wiki.libravatar.org/api/
    _______________________________________________
    Mailing list: https://launchpad.net/~libravatar-fans
    Post to     : libravatar-fans@xxxxxxxxxxxxxxxxxxx
    <mailto:libravatar-fans@xxxxxxxxxxxxxxxxxxx>
    Unsubscribe : https://launchpad.net/~libravatar-fans
    More help   : https://help.launchpad.net/ListHelp



References