<div dir="ltr">On Mon, Sep 30, 2013 at 4:16 PM, Dirk Eddelbuettel <span dir="ltr"><<a href="mailto:edd@debian.org" target="_blank">edd@debian.org</a>></span> wrote:<br><div class="gmail_extra"><div class="gmail_quote">
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><br>
Hi Martyn,<br>
<div class="im"><br>
On 30 September 2013 at 21:02, Martyn Plummer wrote:<br>
| I've already been asked to look into this by Brian. I can't say I've rolled up my sleeves very far, but the problem appears to be in Eigen itself.<br>
<br>
</div>Cool! Good to know.<br>
<div class="im"><br>
| I should clarify that I'm not volunteering to maintain RcppEigen, but I will provide patches as and when.<br>
<br>
</div>That too is appreciated!<br>
<br>
The Eigen code on CRAN has not been updated in a while. The version number<br>
(0.3.1.2.1) reflect both upstream Eigen 3.1.2 (two patch releases behind<br>
Eigen in the 3.1.* series which is now at 3.1.4) and the mandatory change for<br>
JSS once the paper was accepted.<br>
<br>
Changing 3.1.2 to 3.1.3/3.1.4 may help. Or it may not. Maybe 3.2.0 upstream<br>
will help.<br>
<br></blockquote><div>Changing to 3.1.4 or, better, 3.2, would be a good idea. In particular, the 3.2 version provides Cholesky, LU and QR sparse factorizations that do not depend on SuiteSparse (the package of C libraries of which CHOLMOD is part).</div>
<div><br></div><div>Unfortunately, the current maintainer played a little fast and loose by adding a couple of wrinkles to the Eigen code itself and using these wrinkles in lme4.</div><div><br></div><div>I had been hoping to put off the switch to Eigen 3.2 for a while longer as I am facing some deadline pressures on another project. But perhaps it would be best to create an experimental branch using Eigen 3.2. Best case, the fixes for lme4 are easy. Worst case ...</div>
<div><br></div><div>Ben's message came through while I was writing this. If you are interested look at the last link he gives</div><div><br></div><div><a href="http://permalink.gmane.org/gmane.comp.lib.eigen/991">http://permalink.gmane.org/gmane.comp.lib.eigen/991</a></div>
<div><br></div><div>and you will see that the sort of error messages showing up in the RcppEigen log were present in 2010 and the reaction of the Eigen developers is that the compiler is complaining about code that follows the standard and is compilable on other systems.</div>
<div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
All this is now of somewhat greater importance as lme4 on CRAN uses<br>
RcppEigen. "Just make it work and freeze it" is one approach, updating to<br>
current versions is another.<br>
<br>
As I said, it's an open job for someone who wants to earn some karma.<br>
<br>
But in any event, I am glad that you accepted that challenge. Whichever<br>
timeline you can work, it'll be appreciated. We can set you up with R-Forge<br>
commit privileges for this (and thereby the rest of Rcpp) as well.<br>
<br>
Cheers, Dirk<br>
<div class=""><div class="h5"><br>
--<br>
Dirk Eddelbuettel | <a href="mailto:edd@debian.org">edd@debian.org</a> | <a href="http://dirk.eddelbuettel.com" target="_blank">http://dirk.eddelbuettel.com</a><br>
_______________________________________________<br>
Rcpp-devel mailing list<br>
<a href="mailto:Rcpp-devel@lists.r-forge.r-project.org">Rcpp-devel@lists.r-forge.r-project.org</a><br>
<a href="https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel" target="_blank">https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel</a><br>
</div></div></blockquote></div><br></div></div>