<div dir="ltr">I think the pros outweigh the cons -- I'll comment briefly on the PR.<br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, 9 Sep 2019 at 02:41, Matti Picus <<a href="mailto:matti.picus@gmail.com">matti.picus@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div style="direction:ltr" bgcolor="#FFFFFF">
<p>We have discussed using the hypothesis package to generate test
cases at a few meetings informally. At the EuroSciPy sprint,
kitchoi took up the challenge and issued a pull request
<a class="gmail-m_-8562208167951600638moz-txt-link-freetext" href="https://github.com/numpy/numpy/pull/14440" target="_blank">https://github.com/numpy/numpy/pull/14440</a> that actually goes ahead
and does it. While not finding any new failures, the round-trip
testing of s = np.array2string(np.array(s)) shows what hypothesis
can do. The new test runs for about 1/2 a second. In my mind the
next step would be to use this style of testing to expose problems
in the np.chararray routines.<br>
</p>
<p><br>
</p>
<p>What do you think? Is the cost of adding a new dependency worth
the more thorough testing?</p>
<p>Matti<br>
</p>
</div>
_______________________________________________<br>
NumPy-Discussion mailing list<br>
<a href="mailto:NumPy-Discussion@python.org" target="_blank">NumPy-Discussion@python.org</a><br>
<a href="https://mail.python.org/mailman/listinfo/numpy-discussion" rel="noreferrer" target="_blank">https://mail.python.org/mailman/listinfo/numpy-discussion</a><br>
</blockquote></div>