Re: 4.0 release goals

[Jordi Gutiérrez Hermoso]

On 1 November 2012 13:54, Lukas Reichlin <address@hidden> wrote:
> On 01.11.2012, at 17:59, Jordi Gutiérrez Hermoso <address@hidden> wrote:
>
>> I would like to propose some rather ambitious goals for 4.0, and I
>> think we should be bold enough to use a major version number.
>>
>> In case you missed it in the help list, there's an interesting
>> thread about the GUI. Israel Herraiz forked it and stabilised it:
>>
>>    
>> http://octave.1599824.n4.nabble.com/When-will-the-new-GUI-be-ready-tp4645798p4645846.html
>>
>> I would like to prioritise getting the GUI stabilised so we can
>> start contemplating making a release. I think our 4.0 release goals
>> should be:
>>
>>    - fltk enabled by default
>>    - Stable GUI
>>    - JIT compiling working
>>    - Windows and Mac OS X binaries
>>
>> We're already made many inroads into many of these goals. What do
>> you guys think, can we do it?
>>
> Sounds good, but I think Octave should be released early and often.

That is the idea, but early and often was meant to be every six months
or so. We can keep making bug-patching releases of 3.6 for a while,
but we've got some pretty big changes already in the dev branch that
cannot be released right now in whatever state they're in. We simply
recently attempted some really big changes for this release cycle that
can't be broken up into smaller changes to be released piecemeal.

We could just release everything right now as it is, disable the GUI
and JIT compiling by default, and deal with the horrible user
backslash when they get an unstable GUI and have trouble compiling
Octave with JIT compiling, but I don't think this is the best idea.

> If all these features are ready at once, fine.

They're all in various states of development.

> If not, don't hold them back. I propose that the GUI comes with
> Octave 4.0 (as it is the most notable difference). If other features
> like classdef support

classdef is very far away and will not be released any time soon. It's
a horribly complex monster, and we will need lots more geekpower to
get it working.

> or binaries are ready earlier,

Mac OS X binaries have been worked on by Ben. He's done a lot of work
on them.

Windows binaries will probably again need a lot of work, especially
since it seems the current ones don't work on Windows 8 at all.

> create Octave 3.8 or 3.10, 3.12, … If JIT compiling takes longer
> than the GUI to complete, release it as Octave 4.2. Just my 0.02
> SFr.

We'd have to do some repository juggling to release these features
incrementally, since they are all tangled right now into the source
tree. JIT compiling already works right now, but there are still many
things it doesn't compile. It's a pretty innocuous feature that mostly
stays out of the way. The GUI is the biggest priority. Everything else
can released in whatever state it happens to be when the GUI is
stabilised.

My point being, GUI should be the biggest priority, let's stabilise it
so we can make an awesome 4.0 splash.

- Jordi G. H.