Re: [femlisp-user] femlisp update

[Nicolas Neuss]

Faré <address@hidden> writes:

> Dear Nicolas, or whoever else is now maintaining or simply using
> femlisp,

Hi Fare,

it's still me, although I did not work much on Femlisp in the last
years.

> femlisp is unique in binding *read-default-float-format* to
> 'double-float in setup.lisp, and this is causing a bit of a stir as
> I'd like ASDF to bind syntax variables to constant default values
> around each file, which breaks your setup, unless that default is
> 'double-float, which I'd like, but which seems to be the minority view
> on the asdf-devel mailing-list.

Yes, this is not perfect.  Maybe the best way out would be to make
Femlisp accept also single-float if this should be the user's default.
However, I expect that using single-float throughout leads to rather
unusable results due to the ill-conditioning of the stiffness matrix.
Using double-float for some sensitive parts would avoid this, but would
need more work.

> Would you accept a patch?

Yes, if it makes sense.

> While I'm at it, would you move of the CVS repository to github? I can
> do the git conversion. Actually, I just did it:

> https://github.com/fare/femlisp

I don't have anything against switching to git.  However, I would rather
stay with Savannah, because I have some trust in the FSF, and don't know
anything about github apart from that it is a commercial enterprise.

> Please review my 7 commits - some of the "fixes" are only valid up to
> my limited understanding.  

OK, I'll do that.  Anyway, I'll have to reintroduce some changes which
were apparently were lost during some attack on the Savannah server some
years ago.

> You might want to run your tests.  Can you
> take ownership of (a copy of) the project on github?

As said, I would prefer to stay on Savannah where one can open a git
repository as well.  (I am just discussing this on the "savannah-users"
mailing list.)

> I assume ASDF2 or later --- it's been around since 2010, and every
> implementation has adopted it long ago. I refrained from making it
> explicitly depend on ASDF3, though I use some ASDF3 idioms
> conditionally, and assume that implementations known to have adopted
> it already (all but LispWorks have adopted it last year already).

OK, I'll look at that this week, too.

Nicolas