Randomized tests in CppunitTest_basegfx

[Tomaž Vajngerl]

Hi,

On Thu, Feb 23, 2023 at 12:20 AM Michael Meeks <michael.meeks at collabora.com>
wrote:

>         Ooh - interesting; so would using the boxclipper's
> getRandomOrdinal -
> which uses rand() and (I assume) could be consistently seeded ?
>
>         Or could we consistently seed:
>
>         comphelper::rng::uniform_size_distribution(
>
>         To produced a known pattern ? or we could hard-code a set of
> numbers
> (which I guess) wouldn't provoke the problem into the test.
>

Well a unit test should be a set of known input (typical input values and
known corner cases) that asserts a known output, which then defines the
behaviour of the calling function or class. If you have random values for
an input you don't really have a "known" set of input values as they are
generated every time. Would be much better to change this test to a simple
set of  known input values and also tackle the corner cases, which are
missing (but we obviously do hit one).

Also getRandomOrdinal(1) is essentially a NOOP - always returns 0, so the
aPolygonOffset was always offset by 0.5 from the other generated polygon,
which is not what the intended use case for the test was.



>         Would be lovely to find the sequence that kills it of course,
>

It happens because getDeltaY() is 0.0, which is the "end - start" Y value
of an edge, but this one should never be zero (according to the comment).

I have run a simpler case in a loop but didn't get the value 0, so I could
write an edge case unit test. Deconstructing the code and unit test the
parts would probably be needed.


>
>         Thoughts much appreciated,
>
>                 Michael.
>
> --
> michael.meeks at collabora.com <><, GM Collabora Productivity
> Hangout: mejmeeks at gmail.com, Skype: mmeeks
> (M) +44 7795 666 147 - timezone usually UK / Europe
>

Tomaž
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/libreoffice/attachments/20230223/b341c736/attachment.htm>