On 11 December 2005 at 19:24, Steve Langasek wrote:
| On Sun, Dec 11, 2005 at 09:14:00PM -0600, Dirk Eddelbuettel wrote:
| >
| > On 11 December 2005 at 19:02, Steve Langasek wrote:
| > | On Sun, Dec 11, 2005 at 07:49:55PM -0600, Dirk Eddelbuettel wrote:
| > |
| > | > On 11 December 2005 at 17:25, Blars Blarson wrote:
| > | > | Package: foreign
| > | > | Severity: serious
| > | > | Justification: fails to build from source
| > |
| > | > | foreign failed to build on a sparc buildd, duplicated on my sparc
pbuilder.
| > | > | (The number was 27374 rather than 27717 on my pbuilder.)
| > |
| > | > | make[2]: Entering directory
`/build/buildd/foreign-0.8.11/debian/foreign.Rcheck/tests'
| > | > | Running 'S3.R'
| > | > | Comparing 'S3.Rout' to 'S3.Rout.save' ... OK
| > |
| > | > Ok, I'll turn the regression test off as we did with a few other R
packages.
| > |
| > | And ship packages that generate bus errors?!
| >
| > Naaarrr, mate, don't panic. The R packages compare pre-generated output to
| > actual output, and tend to have the numerical precision too tight, in
| > particular for the foreign package that is focussed on importing data sets
| > for other statistical applications.
| >
| > I can detail this some more if you wish but rest assured that there are no
| > bus errors anywhere.
|
| Yes, there are. That was the reason for the failure in the build log that
It only looks that way when the error is caught. A fuller 'R CMD check
foreign' run eg here looks like:
[EMAIL PROTECTED]:~/src/debian/CRAN> R CMD check foreign
* checking for working latex ... OK
* using log directory
'/var/autofs/net/chibud/home/edd/src/debian/CRAN/foreign.Rcheck'
* using R version 2.2.0, 2005-10-06
* checking for file 'foreign/DESCRIPTION' ... OK
* this is package 'foreign' version '0.8-11'
* checking if this is a source package ... WARNING
Subdirectory 'foreign/src' contains object files.
* Installing *source* package 'foreign' ...
checking for gcc... gcc
checking for C compiler default output file name... a.out
[... it now builds the binary yada yada...]
* creating foreign-Ex.R ... OK <-- extracts examples from docs
* checking examples ... OK <-- runs all examples
* checking tests ... <-- starts tests
and I grant you that it _looks_ that way in the log, but it isn't
really.
| was just cited to you. The package also fails to build on ia64 with the
| same bus error.
On the ia64 it also does
* creating foreign-Ex.R ... OK
* checking examples ... OK
* checking tests ...
make[1]: Entering directory
`/build/buildd/foreign-0.8.11/debian/foreign.Rcheck/tests'
showing that binary package is fine as it runs its examples. The generated
file is almost 20k:
[EMAIL PROTECTED]:~/.chibud/src/debian/CRAN> ls -l foreign.Rcheck/*Rout
-rw-r--r-- 1 edd edd 19677 Dec 11 21:39 foreign.Rcheck/foreign-Ex.Rout
"foreign-Ex.Rout" is to be read as $package-Ex.Rout, i.e. R output from
running the examples of package foreign.
I still call it a no-panic situation. Now, as for cdbs, that I call a
panic...
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]
