On 14 October 2006 at 22:31, Martin Michlmayr wrote:
| * Dirk Eddelbuettel <[EMAIL PROTECTED]> [2006-10-14 14:57]:
| > How could that have slipped? Whose shlibs file do we need to look at?
| ...
| > As I will rebuilt all these Rmetrics packages in the next few days (once the
| > new r-cran-fecofin is in the archive), how can I make sure the same problem
| > won't bite us on the move from gfortran 4.1 to 4.2, 4.3, ... and beyond?
|
| I don't know, sorry.
Same here, I'm afraid.
Given how these R packages are built with compiler configuration determined
at the built-time of the R package, and then stored in /etc/R/Makeconf, I
probably need to have R depend on it.
In fact, the 'empty' r-base-dev package does just that, but it has
[EMAIL PROTECTED]:~> apt-cache show r-base-dev | grep gfortran
Depends: r-base-core (>= 2.4.0-1), refblas3-dev | atlas3-base-dev |
libblas-3.so, build-essential, gfortran (>= 4.0.2), gcc (>= 4:4.0), g++ (>=
4:4.0), libncurses5-dev, libreadline5-dev, libjpeg62-dev, libpcre3-dev,
libpng12-dev, zlib1g-dev, libbz2-dev
a 'gfortran (>= 4.0.2)' and it seems to me that gfortran is to blame here if
it does in fact use incompatibale APIs/libs without telling us, no?
Dirk
--
Hell, there are no rules here - we're trying to accomplish something.
-- Thomas A. Edison
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
