[R] nls(() and trace

[Douglas Bates]

joerg van den hoff wrote:
> hi everybody,
> 
> is there a canonical way to get hold of the "trace=TRUE" output from
> nls, i.e. to copy it to a R variable (or at least to an external log file)?
> 
> I have only found the possibility to "fix(nlsModel)" (and than the
> correct copy of that: namespace function ...) within the R-session by
> modifying the trace() definition within nlsModel. not really good for
> everyday use ...
> 
> 
> regards,
> joerg
> 
> 
> ps: if this comes to douglas bates' attention: would'nt it be helpful to
> add the trace output as a further component(possibly optitional,
> depending on the trace flag)  to the nls-object returned by nls()?

My immediate response was to say that it already was but then I checked
the code and found that indeed it is not.  Yes, this is a good idea and
I will put it on the ToDo list but I should warn you that the list is
rather long.  If anyone else feels motivated to do this it would be most
welcome.