octave-maintainers
[Top][All Lists]
Advanced

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

Re: Moving ode changes to default branch


From: Rik
Subject: Re: Moving ode changes to default branch
Date: Fri, 14 Oct 2016 10:51:02 -0700

On 10/14/2016 09:16 AM, c. wrote:
> On 11 Oct 2016, at 17:55, Rik <address@hidden> wrote:
>
>> Similarly, ode23.m has a typo in the docstring:
>> <@qcode{"AbsTol"},.>.  The comma before the period is not grammatically
>> correct.  
> This the change in ode23:
>
> -## computation may be changed by using the option @qcode{"Tau"}, that has a
> -## default value of @math{1e-6}. 
> +## computation may be changed by using the options @qcode{"RelTol"},
> +## and @qcode{"AbsTol"},. 
>
> Indeed an incorrect comma in the documentation is a nuisance and should be 
> fixed.
>
> On the other hand the previous text was wrong and misleading: there is no 
> option called
> Tau and setting such option has no effect.
>
> So, if you revert this change the help text will be wrong and misleading, if 
> you don't
> it will include an inconsistent comma, which one would you prefer to see 
> released?

There's no reason not to fix documentation on the stable branch.  That's
allowed at all times.  Whichever decision is made, the docstring for ode23
needs to be updated.

>
>> With more testing time I'm sure this would be caught.
> Are you sure?

The correlation between length of testing and number of bugs found is
positive.  It doesn't guarantee that any specific bug will be found.  Let
me re-state and say "more confident" this would be caught.  As it is now,
the code has only been in Mercurial for 8 days, with a bug found on day 1. 
That feels like development code, not stable code.

>
> According to 'hg blame' the previous version had been in the repository at 
> least since 
> revision afe9c529760d of december 2015 and had gone unnoticed ...

"If a tree falls in the forest, does it make a sound?"  If the code is
never used, does it matter that there are bugs in it?  Apparently no one
used this feature of ode23 so maybe we shouldn't be that concerned.  In an
ideal world, all Octave functions would have tests and there would be 100%
line coverage, but the team is pretty small.

To move forward, could you, Jacopo, or anyone else who has a strong
interest in the ode functions do a code review and post the changes to the
patch tracker.  From the magnitude and scope of the changes we can derive a
better idea of whether this can remain on stable.

--Rik




reply via email to

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