AW: fUnitRoots failing with urca import due to BIC

[Martin Maechler]

>>>>> Dirk Eddelbuettel <edd at debian.org>
>>>>>     on Tue, 7 Jun 2011 08:51:49 -0500 writes:

    > Hi Bernhard,

    > Thanks for the note.

    > On 30 May 2011 at 08:25, Pfaff, Bernhard Dr. wrote: |
    > Hello Dirk,
    > | 
    > | a couple of weeks ago BDR reminded package maintainers
    > of a change with respect to BIC. He suggested for 'urca'
    > the NAMESPEACE insertion you provided below. The latest
    > CRAN Package Check does not indicate an error for
    > 'urca'. If I recall correctly, the package fUnitRoots was
    > mentioned in that email of BDR, too. Maybe the required
    > changes have not implemented in there?

    > Seems like it, and the R-Metrics team appears to be too
    > busy before the upcoming retreat.

    > Would you have a suggestion as to how to fix this on the
    > side of fUnitRoots so that it builds again?  Else I may
    > have to withdraw the package.

which one? (:-)

Note that  fUnitRoots  does not once use  BIC in its code,
i.e. the error really happens when  require("urca") is used 
somewhere.

Also, the error message below indicates a problem with urca
(only)?

Interestingly, when I tried things, I got the same error as the
Debian testing machine.

But then I re-installed  urca  and all was fine.

So, I hope that this (first re-install urca, then check and
install fUnitRoots) should also work on the debian test
machine(s) ?

Martin

 | 
 | Best,
 | Bernhard   
 | 
 | > -----Ursprüngliche Nachricht-----
 | > Von: Dirk Eddelbuettel [mailto:edd at debian.org] 
 | > Gesendet: Samstag, 28. Mai 2011 23:53
 | > An: rmetrics-core; Pfaff, Bernhard Dr.
 | > Betreff: fUnitRoots failing with urca import due to BIC
 | > 
 | > 
 | > Hi all,
 | > 
 | > During a routine rebuild of the Debian archive, it was 
 | > noticed that fUnitRoots no longer builds 
 | > (http://bugs.debian.org/628324):
 | > 
 | > make[1]: Entering directory `/tmp/buildd/funitroots-2100.76/src'
 | > gfortran   -fpic  -O3 -pipe  -g -c UnitRootMacKinnon.f -o 
 | > UnitRootMacKinnon.o
 | > gcc -shared -o fUnitRoots.so UnitRootMacKinnon.o -llapack 
 | > -lblas -lgfortran -lm -lquadmath -lgfortran -lm -lquadmath 
 | > -L/usr/lib/R/lib -lR
 | > make[1]: Leaving directory `/tmp/buildd/funitroots-2100.76/src'
 | > make[1]: Entering directory `/tmp/buildd/funitroots-2100.76/src'
 | > make[1]: Leaving directory `/tmp/buildd/funitroots-2100.76/src'
 | > installing to 
 | > /tmp/buildd/funitroots-2100.76/debian/r-cran-funitroots/usr/li
 | b/R/site-library/fUnitRoots/libs
 | > ** R
 | > ** inst
 | > ** preparing package for lazy loading
 | > Error : object 'BIC' is not exported by 'namespace:nlme'
 | > Error : package 'urca' could not be loaded
 | > ERROR: lazy loading failed for package 'fUnitRoots'
 | > 
 | > 
 | > Did something change in nlme which upsets urca?  I only see (in urca's
 | > NAMESPACE)
 | > 
 | > ## Import functions
 | > 
 | > importFrom(graphics, plot)
 | > 
 | > if (!exists("BIC", envir=asNamespace("stats"))) {
 | >     importFrom(nlme, BIC)
 | > } else {
 | >     importFrom(stats, BIC) # unless you already import stats }
 | > 
 | > I tried adding 'stats' to the Depends: in urca's DESCRIPTION 
 | > but not beans. 
 | > 
 | > Any ideas?
 | > 
 | > --
 | > Gauss once played himself in a zero-sum game and won $50.
 | >                       -- #11 at http://www.gaussfacts.com