R CMD javareconf in r-base-dev 2.6.0

Hi Johannes,

| Hi Dirk, hi list,
| 
| I just ran into a slight problem while building the recommended packages
| for CRAN in a chroot environment against the fresh backport of your R
| 2.6.0 packages. It seems that R CMD javareconf wants to change files in

You are ahead of me, as this hook is to be used by rJava and CRAN packages
using rJava.  I actually haven't used it yet, as I didn't get around to
putting it in, and the testing it. So far only step 1 has been done :)

[ The idea is to have the ability of building jar archvies using tools from
Debian main.  Users can then switch to Sun's or IBM's Java (which the buildd
cannot depend on for lack of freeness) and things still work, according to
Simon who knows R and Java inside out. ]

| /etc/R during the build, and since I am not building the packages as
| root in my etch chroot, this fails, even if I am using fakeroot. My

Hm, I am confused.  Do you build manually in the chroot?  Or do you use
pbuilder?  I don't think my pbuilder has gij and hence /usr/bin/java.  But
then you are rebuilding for testing so things ought to be different....

| workaround was to simply remove gij-4.1 from the chroot, so
| /usr/bin/java is not found any more.

Sounds like exactly the right solution.  Hope it is not too much of a bother.

Dirk

Three out of two people have difficulties with fractions.

Hi Dirk,

* Dirk Eddelbuettel <edd at debian.org> [071003 23:50]:

Hi Johannes,

On 3 October 2007 at 22:23, Johannes Ranke wrote:
| Hi Dirk, hi list,
| 
| I just ran into a slight problem while building the recommended packages
| for CRAN in a chroot environment against the fresh backport of your R
| 2.6.0 packages. It seems that R CMD javareconf wants to change files in

You are ahead of me, as this hook is to be used by rJava and CRAN packages
using rJava.  I actually haven't used it yet, as I didn't get around to
putting it in, and the testing it. So far only step 1 has been done :)

I wonder if it is necessary to run this command (if /usr/bin/java
exists) for every R package that is being built, regardless if it needs
java or not. I have no idea how many people are out there building debs
for R packages, and you'd probably argue that they better use pbuilder.

[ The idea is to have the ability of building jar archvies using tools from
Debian main.  Users can then switch to Sun's or IBM's Java (which the buildd
cannot depend on for lack of freeness) and things still work, according to
Simon who knows R and Java inside out. ]

This sounds very nice indeed.

| /etc/R during the build, and since I am not building the packages as
| root in my etch chroot, this fails, even if I am using fakeroot. My

Hm, I am confused.  Do you build manually in the chroot?  Or do you use
pbuilder?  I don't think my pbuilder has gij and hence /usr/bin/java.  But
then you are rebuilding for testing so things ought to be different....

Yes, I am using my own script to build in a "stable" chroot without
pbuilder so far.

| workaround was to simply remove gij-4.1 from the chroot, so
| /usr/bin/java is not found any more.

Sounds like exactly the right solution.  Hope it is not too much of a bother.

Not at all, I just thought it might be interesting for you to hear.
 
Johannes

Dirk

-- 
Three out of two people have difficulties with fractions.

Dr. Johannes Ranke                 jranke at uni-bremen.de     Key ID: F649AF90
UFT Bremen, Leobenerstr. 1         +49 421 218 63373
D-28359 Bremen                     http://www.uft.uni-bremen.de/chemie/ranke

Hi Johannes,

Many thanks for the speedy built of the backports from my R 2.6.0 package!

| Hi Dirk,
| 
| * Dirk Eddelbuettel <edd at debian.org> [071003 23:50]:
| > 
| > Hi Johannes,
| >

| > | Hi Dirk, hi list,
| > | 
| > | I just ran into a slight problem while building the recommended packages
| > | for CRAN in a chroot environment against the fresh backport of your R
| > | 2.6.0 packages. It seems that R CMD javareconf wants to change files in
| > 
| > You are ahead of me, as this hook is to be used by rJava and CRAN packages
| > using rJava.  I actually haven't used it yet, as I didn't get around to
| > putting it in, and the testing it. So far only step 1 has been done :)
| 
| I wonder if it is necessary to run this command (if /usr/bin/java
| exists) for every R package that is being built, regardless if it needs
| java or not. I have no idea how many people are out there building debs
| for R packages, and you'd probably argue that they better use pbuilder.

Well, but are other users really calling r-cran.mk ?  

To the best of my knowledge, it is used only in the (numerous) r-cran-*
packages --- which are all built in a pbuilder chroots that do _not_ contain
/usr/bin/java by default so the test comes up false and no R CMD javareconf
is run --- and the (not really official) pkg-bioc effort of automatic
creation of Debian package, which also uses pbuilder behaves the same. 
So I saw no downside.

What have I overlooked?  If and when users install packages directly from
CRAN via R's install.packages(), this command isn't run either.

But I may have overlooked something obvious, so let me know. We can always
back it out for a -2 revision.

Cheers, Dirk

Three out of two people have difficulties with fractions.

Hi Dirk,

* Dirk Eddelbuettel <edd at debian.org> [071004 14:40]:

Hi Johannes,

Many thanks for the speedy built of the backports from my R 2.6.0 package!

It's easy, basing on your work.

On 4 October 2007 at 08:30, Johannes Ranke wrote:
| Hi Dirk,
| 
| * Dirk Eddelbuettel <edd at debian.org> [071003 23:50]:
| > 
| > Hi Johannes,
| > 
| > On 3 October 2007 at 22:23, Johannes Ranke wrote:
| > | Hi Dirk, hi list,
| > | 
| > | I just ran into a slight problem while building the recommended packages
| > | for CRAN in a chroot environment against the fresh backport of your R
| > | 2.6.0 packages. It seems that R CMD javareconf wants to change files in
| > 
| > You are ahead of me, as this hook is to be used by rJava and CRAN packages
| > using rJava.  I actually haven't used it yet, as I didn't get around to
| > putting it in, and the testing it. So far only step 1 has been done :)
| 
| I wonder if it is necessary to run this command (if /usr/bin/java
| exists) for every R package that is being built, regardless if it needs
| java or not. I have no idea how many people are out there building debs
| for R packages, and you'd probably argue that they better use pbuilder.

Well, but are other users really calling r-cran.mk ?  

To the best of my knowledge, it is used only in the (numerous) r-cran-*
packages --- which are all built in a pbuilder chroots that do _not_ contain
/usr/bin/java by default so the test comes up false and no R CMD javareconf
is run --- and the (not really official) pkg-bioc effort of automatic
creation of Debian package, which also uses pbuilder behaves the same. 
So I saw no downside.

What have I overlooked?  If and when users install packages directly from
CRAN via R's install.packages(), this command isn't run either.

But I may have overlooked something obvious, so let me know. We can always
back it out for a -2 revision.

I see no real problem at the moment either. Let's see what happens when
you actually package R packages using java!

Best,

Johannes

Cheers, Dirk

-- 
Three out of two people have difficulties with fractions.