[Rd] glm start/offset bugs (PR#1422)

Brett Presnell presnell@stat.ufl.edu
Fri, 29 Mar 2002 11:51:19 -0500

Prof Brian D Ripley writes:
> Yes, thought has been given, and the code has been looked at. (People
> do look at R code you know, even if they don't jump in to change it
> after a few hour's use.)

Sorry. No insult intended.

> Given the time pressures, this is not going to get looked at for
> 1.5.0.

Does this mean that my patch will not make it in either?  Berwin's
work is obviously more complicated and requires more attention and
careful reading to check.  My changes actually only have an effect if
both an offset and a starting value are given, and clearly the current
version falls down badly in that case anyway.

> I have long thought that given the history of problems maintaining
> glm() that it might be easier to rewrite it for scratch.  Having
> .null versions serves no useful purpose, for example.

Yes, having ploughed around in the code a bit, I'm very sympathetic to
this view.

Brett Presnell
Department of Statistics
University of Florida
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch