Re: problem with failing unit test in stk : any ideas ?

[Rik]

On 11/01/2015 12:53 AM, address@hidden wrote:

Subject:

Re: problem with failing unit test in stk : any ideas ?

From:

Julien Bect <address@hidden>

Date:

10/31/2015 11:58 AM

To:

address@hidden

List-Post:

<mailto:address@hidden>

Content-Transfer-Encoding:

quoted-printable

Precedence:

list

MIME-Version:

1.0

References:

<address@hidden> <address@hidden> <address@hidden> <address@hidden>

In-Reply-To:

<address@hidden>

Message-ID:

<address@hidden>

Content-Type:

text/plain; charset=utf-8; format=flowed

Message:

1

Le 31/10/2015 18:03, Mike Miller a écrit :

On Sat, Oct 31, 2015 at 17:42:48 +0100, Julien Bect wrote:

>So, if I understand correctly your parable, I'm not supposed to write this
>sort of test... How am I supposed to know that? What is wrong with my test?

I think Rik's point was that it turns out to be a corner case and may
not be worth fixing versus working around. Not that it was supposed to
be obvious to not do in the first place :)

Well... if you think that "sticking a knife in one's side" is a metaphor for "corner case"... perhaps ;-)

Mike has summarized my position accurately.  Given the apparently low incidence rate of this bug I would choose to avoid it rather than solve it.  I am speaking of my own coding priorities, but if this "itch" is really bothersome then the beauty of open source is that anyone is free to spend time scratching it.
 

My opinion is that this seemingly unimportant problem might indicate some deeper issue with eval / function / clear.

There is no doubt that something is wrong, but it can't be too awful because eval does work most of the time.  The core uses it extensively to run the BIST tests, there are a few tests in oct-parse.cc, and a longer set of tests in eval-catch.tst.  I expect we would also have loads of bug reports about eval if it were badly broken.

I have tried to isolate what test does and filed a bug report: https://savannah.gnu.org/bugs/?46350.

This is good.  An unambiguous repeatedly failing case is a great help for debugging.

--Rik