Re: Policy of refering to packages? Policy of choosing package for ML-compatible functions?

[PhilipNienhuis]

Olaf Till-2 wrote
> Dear maintainers,
> 
> as mentioned in a recent post at the help list
> 
> http://lists.gnu.org/archive/html/help-octave/2016-12/msg00134.html
> 
> Octave may refer to the wrong OF package for functions not in core. It
> seems that the reference only considers in which ML toolbox the
> corresponding function is implemented.
> 
> I could just file a bug, but I think it would be better to discuss and
> decide on a policy for these references. Problems could be that a
> function may not yet be implemented even in OF and that a function may
> be moved from one package to a different one.
> 
> This can't be separated from the question if a ML-compatible function
> should always be contained in a package corresponding to the
> respective ML-toolbox. So I think this issue should also be discussed
> in this context. This had been attempted by Julien in
> 
> http://lists.gnu.org/archive/html/octave-maintainers/2015-08/msg00100.html
> 
> but there was no discussion.

Well there was a discussion at the time of the polygon functions: i.e.,
whether they should go in OF-mapping, or in OF-geometry:  see here.
<http://octave.1599824.n4.nabble.com/Re-regarding-location-of-Matlab-compatible-polygon-functions-tt4676164.html#none>
  

My stance there was (and is) that IMO "Matlab compatibility" is primarily
code compatibility and not so much striving for verbatim copies of Matlab's
ecosystem.

In principle it isn't such a big deal as to what package contains which
functions. After all, unlike with Matlab, no money is involved when
installing additional packages ("toolboxes"), isn't it. For a simple "pkg
install -forge....." no credit card details are required.

But package names do raise expectations about contents.
In the case of OF-mapping vs. OF-geometry referenced above an important (for
me) consideration was that Matlab does not have a package dedicated to
geometry operations while Octave does.
So should we then cripple the OF-geometry package by placing functions
naturally belonging there into some other package just because of what
Matlab users expect?

My thinking is that it is up to the package maintainer(s) to decide what
functionality goes where. The trade-off between e.g., "matlab users
expectations" versus different Octave package goals and contents is one to
be made by the package maintainers, in the end, after discussion.

Philip




--
View this message in context: 
http://octave.1599824.n4.nabble.com/Policy-of-refering-to-packages-Policy-of-choosing-package-for-ML-compatible-functions-tp4681119p4681120.html
Sent from the Octave - Maintainers mailing list archive at Nabble.com.