[ Thread Index |
| More lists.tuxfamily.org/eigen Archives
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] BTL
- From: Andrea Arteaga <yo.eres@xxxxxxxxx>
- Date: Mon, 5 Sep 2011 23:00:43 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:sender:in-reply-to:references:from:date :x-google-sender-auth:message-id:subject:to:content-type; bh=Adf2DlDvQ6hz9V75hXM933UUJlJN5vKWma/cyPrM+i4=; b=q3CjP1azu3VGINT6aHteeIJDJGvapUF6XjCvvlhEdQpibyuHwn0LNOhwy4spA27WuX aB0Ek5fyMHRsJKhHzMkKloJ5VfJgu5jxvR9cGIk+GckzcpyLW9clZbV2EO9j6jcSbBh+ X2EB2pPB2FQQ73LHWauj2latqbEgBT/6SyhUI=
Thanks for your feedback, Gael.
Actually, you are right about the dynamic loading issue: it could be
really difficult to make it portable. Glib could help, but I don't
know how really portable it would be and anyway this would add an
external dependency, which is not a tragic thing, but...
However, this only regards some interfaces: we can provide interfaces
which dynamically load blas and/or lapack shared objects, and the user
*can* decide to use these -- if his/her system is able to do that --,
but we also can provide "standard" blas/lapack interfaces, like the
already existing ones. So, I would say that for now, I/we write the
core entities, then provide the most important actions and interfaces
and at a later time we can start and really think about this
dynamically loading possibility.