Re: [Quantlib-dev] May you give us an overview of current QuantLib plans?

[Luigi B. <lui...@gm...>]

Hi Piter,

On Mon, 2009-08-10 at 07:16 -0300, Piter Dias wrote:
> I know there is plan to deliver QuantLib 1.0 as a stable and
> production version. In fact, I believe we already has production users
> of QuantLib, but I believe that whole idea is the project itself would
> change its status to a Production software with this release.

Yes. In particular, that would mean that we'll ensure backward
compatibility of future releases with version 1.0, meaning that code
that compiles against 1.0 will compile unchanged against later versions
(with an exception---see later.)  This is also the reason for the delay
in getting a new release out: we needed time to fix a few things before
freezing the interfaces.


> However, it also means that some contributions would have to wait more
> in order to not cause a mis function in the 1.0 version.

Not necessarily. See later.


> May you give us (user and eventual contributor) an overview about what
> your are doing and current plans? I have myself some questions:
> Are you already have a desirable release date? I mean some data you
> wish to deliver but doesn't know if  you will be able to?

We're almost there.  There's still a couple of things to fix, but I'd
like to create the release branch soon and release a version 0.9.9 in
the second half of September (considering that I'll be on vacation next
week and half of the following.)  Unless anything big and unexpected
happens, we should be able to stick to the schedule.  The 0.9.9 version
will practically be a beta version of 1.0; we can expect the latter for
October or November.


> If new  classes and functions  are not supposed  to go to trunk as
> fast as it was, how could we keep people motivation (everyone that
> contributes good stuff want to see it code in the library, right)? Are
> you more open to include test suite contributions to this new release?

This release is almost frozen, so new contributions will go after
version 1.0.  However, that doesn't mean that they'll have to wait long
to be included; once the 0.9.9 and 1.0 branches are created (in a couple
of weeks) new code can go on the trunk for future inclusion in the 1.1
release.  I hope to get back to a decent release schedule (say, 3 or 4
releases per year) so that contributions don't have to wait much to get
out.


> What is the experimental folder really is? Would you deliver it in the
> QuantLib 1.0? What the polices to put or delete code there are?

The experimental folder is an exception to the backward compatibility;
what's in there might change interface in the future. The reason is that
on the one hand, we want contributions to get out; but on the other
hand, we don't want to freeze interfaces we barely used before having
some feedback. New stuff is going to go in experimental first and be
moved in the core folders once we're happy with the interface.


> We have a couple of examples and hundred of tests. Do you believe that
> it is a problem for new users? I mean, I already contribute with test
> but never though about examples. Do you expect more examples in the
> library or test suite itself is THE example?

I think the test suite can work as an example too; we don't need to
write examples to show functionality that is covered in the test suite
(I'd rather add some comments to the test cases.)  Examples might be
used to show more complex workflows.


I'll pass on the QuantLibXL questions, as I'm not very involved in the
project at this time...

Hope this helped. If anybody has any feedback, I'll be glad to hear it.

Luigi


-- 

Olmstead's Law: 
After all is said and done, a hell of a lot more is said 
than done.