Re: [Octave Forge] Octave 4.0 call for packages

[Philip Nienhuis]

Mike Miller-4 wrote
> On Tue, Apr 7, 2015 at 10:46:38 -0700, Philip Nienhuis wrote:
>> Looking at general, I wonder if it would be a lot of work to simply drop
>> the
>> inputParser stuff, add a dependency on Octave-4.0.0+ and then make a new
>> release? (bumping just the last digit of the version number)
>> I've only got general pkg installed because it was/is a dependency for
>> several other packages.
> 
> Personally I would prefer to keep inputParser in general and allow it
> to shadow the one in Octave until 4.0 has been out for a while, for
> some definition of "a while". It's safer, allows people to upgrade
> more smoothly, allows users to stick with Octave 3.8 but update Forge
> packages, etc. But if the rest of you would rather remove it and put
> out a new general depending on Octave 4.0, don't let me hold you back.

If I correctly understand what Carnë wrote to me in private discussion is
that inputParser was already outdated for quite some time.
People desperate to install the old inputParser can always download &
install older versions from octave-forge.


>> Similar reasoning might hold for the specfun package; I believe I've seen
>> the shadow warnings about the ellipke, expint and erfcinv functions for 3
>> or
>> 4 years now (maybe even longer).Do these functions really add something
>> to
>> the core functions?
>> The specfun package functions expint_E1 and expint_Ei work fine with core
>> expint().
> 
> I imported these functions into core for 3.8, the specfun versions do
> not add anything. 

What purpose do expint_E1 and expint_Ei have? Is their functionality
somewhere else in core?


> The same logic as above applies, they still exist in
> specfun because no one has made a new release of the package to remove
> those functions since being added to core. I mentioned earlier in
> another thread that I would be in favor of dropping the entire specfun
> package now.

The specfun package contains several functions that appear to be useful. 
I'm not against a new release w/o the shadowing functions as long as the
rest isn't harmful.


> I believe you touched on the difficulty of dropping complete packages
> in another thread, users are still under the impression that they have
> to install the java package for example. Any ideas for how to better
> inform users that they don't need to install java or specfun anymore?
> Put out a new release of the package with no contents?

Just a new release with appropriate dependencies in DESCRIPTION and NEWS
update, no other changes:

- java package dependent on Octave <= 3.6.4

- specfun <= Octave-3.8.2 as per your suggestion. But I'm not sure I agree
:-)

In a general sense I think it may be a good idea to start adding maximum
Octave version dependencies for all packages (say, <= 4.2.0 for now), and
for each new intermediate Octave release evaluate what changes are required
to run each OF package in that new Octave release. That way the risk of
unwary users installing packages that break things can fairly effectively be
diminished.
Yes it'll place a burden on package maintainers :-)

Philip




--
View this message in context: 
http://octave.1599824.n4.nabble.com/Octave-Forge-Octave-4-0-call-for-packages-tp4669204p4669673.html
Sent from the Octave - Maintainers mailing list archive at Nabble.com.