Re: problems with optimset option name completion in 3.8.0

[Olaf Till]

On Mon, Jan 06, 2014 at 01:51:08PM -0500, Mike Miller wrote:
> On Mon, Jan 6, 2014 at 19:31:22 +0100, Olaf Till wrote:
> > On Fri, Jan 03, 2014 at 11:34:46PM +0100, Olaf Till wrote:
> >> ...
> >> 1) consider it illegitimate to have option names whose first
> >>    characters are equal to a different full option name (means I have
> >>    to change the option names in the said functions of the optim
> >>    package!),
> >>
> >> 2) change optimset so that it always matches an option name if it is
> >>    given in full, even if there is a longer option name which starts
> >>    with the same characters,
> >>
> >> 3) drop option name completion in optimset.
> >> ...
> >
> > No response means I have to go ahead with 1).
> 
> Hi Olaf, I also meant to reply to this but forgot.
> 
> IMHO #2 seems like the appropriate fix for this, if I'm understanding
> you right. If an option name matches exactly it shouldn't complain
> about ambiguity. I can't say whether this is Matlab-compatible or not.

Mmh, I'm not so sure. This could lead to the error that users try to
set the option with the longer name, abbreviating it with the shorter
full option name, and so rather set the option with the shorter name
without noticing.

> Doesn't your given example demonstrate that optimset does set the
> named option appropriately despite showing a warning? I wouldn't
> consider that the worst possible behavior to have to put up with for
> now.

I wouldn't like the warning, and even would not like to have to switch
it off, once this should be possible ... But no warning would lead to
the error possibility I just mentioned ...

Regards, Olaf

> It would also be an improvement if the warnings in optimset had
> warning-ids so they could be disabled individually.
> 
> -- 
> mike

-- 
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net

signature.asc
Description: Digital signature