Re: Avoid some crashes for bad "control-points" property (issue 576610043 by address@hidden)

[David Kastrup]

address@hidden writes:

> On 2019/04/26 18:49:03, dak wrote:
>> mailto:address@hidden writes:
>
>> > add a regression test?
>> >
>> > https://codereview.appspot.com/576610043/
>
>> I am not sure what the point of it would be.  This kind of bug may be
>> present elsewhere (and the regression test would not cover that) and
> it
>> is not likely to get reintroduced here since this code does not really
>> interact with other code.  And if you try running the test on older
>> versions for comparison, it will just crash.  The visual output is
>> meaningless since the condition it protects against does not have
>> meaning.
>
> After many years of developing software, I'm skeptic of any bugfix
> without test, but yes, maybe it's overkill in this case? In any case, I
> think there is a case in slur-configuration.cc too

You are right, and I'll have to fix that.  Seems my grep-fu is
lacklustre.  But a regression test would not have found it.  This kind
of bug seems more amenable to review.  Regression tests would protect
against badly resolved merge conflicts reintroducing old code, of
course, but our workflow does not really make them probable: there are
just too few people working on too few things.

I mean, feel free to propose a regtest.

> https://codereview.appspot.com/576610043/

-- 
David Kastrup