octave-maintainers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Failures in 'make check'


From: Carnë Draug
Subject: Re: Failures in 'make check'
Date: Thu, 6 Oct 2016 23:43:56 +0100

On 6 October 2016 at 21:55, c. <address@hidden> wrote:
>
> On 6 Oct 2016, at 22:04, Carnë Draug <address@hidden> wrote:
>
>> On 6 October 2016 at 19:14, Rik <address@hidden> wrote:
>>> The test suite is failing quite badly because of some changes is the ode
>>> functions.
>>>
>>> ***** test  # two output arguments
>>> [t, y] = ode23 (@fpol, [0 2], [2 0]);
>>> assert ([t(end), y(end,:)], [2, fref], 1e-3);
>>> !!!!! test failed
>>> 'args' undefined near line 45 column 21
>>>
>>> Could someone who knows about this take  look.
>>>
>>
>> I too am seeing 25 and 29 test failures in ode23 and ode45 respectively.
>> Should we be making changes that completely rewrite the internals of
>> functions, fix no bug, in the stable branch, and after two release
>> candidates?  This changes seem better suited to the default branch
>> where it can be tested and reviewed before a release.
>>
>> Carnë
>>
>
> [....]
>
> Summary:
>
>   PASS       100
>   FAIL         0
>

Good that you have fixed it now but that is not the point.

These changes [1, 2] completely rewrite odeget and odeset, around 1500
lines of code get changed.  Changes of this nature should have not been
gone in stable at all, specially after 2 release candidates, an imminent
4.2 release, and no review.

We all want the next release to have as many features as possible but we
also want that release to work properly which is why there is a release
process.  This changes do not respect that process.

We had tests that caught this right away but tests can only catch bugs for
the cases they cover.  There's a default branch for that, where things are
known to break and many of our users test their code there.  This changes
should have gone through the same process.

Note that this is true even if the changes did not cause any test to fail.
Should this changes be reverted from stable?

Carnë

[1] http://hg.savannah.gnu.org/hgweb/octave/rev/dba5074bdc79
[2] http://hg.savannah.gnu.org/hgweb/octave/rev/b8d525710075



reply via email to

[Prev in Thread] Current Thread [Next in Thread]