Hi Tom, what are your plans for R-2.7.0 that was released today? FWIW I am asking just for curiosity. :-)
Regards,
On Tue, 2008-04-22 at 17:28 +0100, José Matos wrote:
Hi Tom, what are your plans for R-2.7.0 that was released today? FWIW I am asking just for curiosity. :-)
I was going to look through the changelog and try to determine whether it merited a last minute tag for F-9 or not. Opinions welcome.
~spot
On Tue, Apr 22, 2008 at 01:37:40PM -0400, Tom spot Callaway wrote:
On Tue, 2008-04-22 at 17:28 +0100, José Matos wrote:
Hi Tom, what are your plans for R-2.7.0 that was released today? FWIW I am asking just for curiosity. :-)
I was going to look through the changelog and try to determine whether it merited a last minute tag for F-9 or not. Opinions welcome.
I don't know about this particular case, but unless it fixes a known bug, I don't like last minute tags, I think the rawhide testing should not be bypassed.
Only my opinion, and since I only run rawhide, no need to weigth it too much.
-- Pat
On Tuesday 22 April 2008 18:37:40 Tom "spot" Callaway wrote:
I was going to look through the changelog and try to determine whether it merited a last minute tag for F-9 or not. Opinions welcome.
AFAIK there is nothing so serious that can't wait for F-9 release. Also all modules need to be recompiled, this can be done best after F-9.
Since the release schedule of R is 6 months I think that we should continue the policy, that has been followed, to release the new version as soon as feasible for all releases. This is different from the procedure used in perl, python, tcl but I think that in this case it is justified. Just my 2 ¢ (euro) on this issue. :-)
~spot
José Matos wrote:
On Tuesday 22 April 2008 18:37:40 Tom "spot" Callaway wrote:
I was going to look through the changelog and try to determine whether it merited a last minute tag for F-9 or not. Opinions welcome.
AFAIK there is nothing so serious that can't wait for F-9 release. Also all modules need to be recompiled, this can be done best after F-9.
I need to recompile rkward against the new R, but this can be done via devel and the testing done via the testing repo.
Fine for me :)
Pierre
"JM" == José Matos writes:
JM> On Tuesday 22 April 2008 18:37:40 Tom "spot" Callaway wrote:
I was going to look through the changelog and try to determine whether it merited a last minute tag for F-9 or not. Opinions welcome.
JM> AFAIK there is nothing so serious that can't wait for F-9 JM> release. Also all modules need to be recompiled, this can be done JM> best after F-9.
I agree, wait for F-9. Then we can put it through updates-testing, have time to push dependent packages that might need rebuilding, including (but probably not limited to) rpy.
Meanwhile we could push versions for F-7 & F-8 to updates-testing to shake out potential problems. Then push them all to stable after F-9 is released.
Alex
On Wed, 2008-04-23 at 02:56 -0700, Alex Lancaster wrote:
I agree, wait for F-9. Then we can put it through updates-testing, have time to push dependent packages that might need rebuilding, including (but probably not limited to) rpy.
Meanwhile we could push versions for F-7 & F-8 to updates-testing to shake out potential problems. Then push them all to stable after F-9 is released.
Ehhh, i don't want to break the upgrade path. We'll just sit on this until after F-9.
~spot
"TC" == Tom "spot" Callaway writes:
TC> On Wed, 2008-04-23 at 02:56 -0700, Alex Lancaster wrote:
I agree, wait for F-9. Then we can put it through updates-testing, have time to push dependent packages that might need rebuilding, including (but probably not limited to) rpy.
Meanwhile we could push versions for F-7 & F-8 to updates-testing to shake out potential problems. Then push them all to stable after F-9 is released.
TC> Ehhh, i don't want to break the upgrade path. We'll just sit on TC> this until after F-9.
I just noticed that some packages fail to rebuild with R 2.7.0, e.g.
http://koji.fedoraproject.org/koji/taskinfo?taskID=606646
the build.log has the following:
checking for unpackaged file(s): /usr/lib/rpm/check-files /var/tmp/R-zoo-1.5-3.fc10-root-mockbuild error: Installed (but unpackaged) file(s) found: /usr/share/R/library/zoo/NEWS RPM build errors: Installed (but unpackaged) file(s) found: /usr/share/R/library/zoo/NEWS
Does 2.7.0 somehow attempt to compile/provide a NEWS file even if the actual package does not?
Alex
"AL" == Alex Lancaster writes:
"TC" == Tom "spot" Callaway writes:
TC> On Wed, 2008-04-23 at 02:56 -0700, Alex Lancaster wrote:
I agree, wait for F-9. Then we can put it through updates-testing, have time to push dependent packages that might need rebuilding, including (but probably not limited to) rpy.
Meanwhile we could push versions for F-7 & F-8 to updates-testing to shake out potential problems. Then push them all to stable after F-9 is released.
TC> Ehhh, i don't want to break the upgrade path. We'll just sit on TC> this until after F-9.
AL> I just noticed that some packages fail to rebuild with R 2.7.0, AL> e.g.
AL> http://koji.fedoraproject.org/koji/taskinfo?taskID=606646
AL> the build.log has the following:
AL> checking for unpackaged file(s): /usr/lib/rpm/check-files AL> /var/tmp/R-zoo-1.5-3.fc10-root-mockbuild error: Installed (but AL> unpackaged) file(s) found: /usr/share/R/library/zoo/NEWS RPM build AL> errors: Installed (but unpackaged) file(s) found: AL> /usr/share/R/library/zoo/NEWS
AL> Does 2.7.0 somehow attempt to compile/provide a NEWS file even if AL> the actual package does not?
Actually looking at it a bit closer it seems that R 2.7.0 is a bit smarter about installing the NEWS file if it's present, see:
https://bugzilla.redhat.com/show_bug.cgi?id=445635#c11
for more details. This might affect a few packages that don't explicitly include NEWS in their %files list. If this is so, it would probably be useful if the sample spec files in the R packaging guidelines:
http://fedoraproject.org/wiki/Packaging/R
were also updated accordingly, as many people use those as templates to create new R packages.
Alex
On Tue, 2008-05-13 at 03:01 -0700, Alex Lancaster wrote:
This might affect a few packages that don't explicitly include NEWS in their %files list. If this is so, it would probably be useful if the sample spec files in the R packaging guidelines:
Good catch, I've updated the guidelines.
~spot
Tom "spot" Callaway wrote:
On Wed, 2008-04-23 at 02:56 -0700, Alex Lancaster wrote:
I agree, wait for F-9. Then we can put it through updates-testing, have time to push dependent packages that might need rebuilding, including (but probably not limited to) rpy.
Meanwhile we could push versions for F-7 & F-8 to updates-testing to shake out potential problems. Then push them all to stable after F-9 is released.
Ehhh, i don't want to break the upgrade path. We'll just sit on this until after F-9.
What about pushing it now ? :)
I have started to look at the new bioconductor release, some of them will require R >= 2.7.0.
Cheers
Pierre
On Sat, 2008-05-17 at 13:23 +0200, pingou wrote:
Tom "spot" Callaway wrote:
On Wed, 2008-04-23 at 02:56 -0700, Alex Lancaster wrote:
I agree, wait for F-9. Then we can put it through updates-testing, have time to push dependent packages that might need rebuilding, including (but probably not limited to) rpy.
Meanwhile we could push versions for F-7 & F-8 to updates-testing to shake out potential problems. Then push them all to stable after F-9 is released.
Ehhh, i don't want to break the upgrade path. We'll just sit on this until after F-9.
What about pushing it now ? :)
I have started to look at the new bioconductor release, some of them will require R >= 2.7.0.
It is now in testing for all active releases (F-7, F-8, F-9):
https://admin.fedoraproject.org/updates/F7/pending/rpy-0.4.6-25.fc7,R-2.7.0-... https://admin.fedoraproject.org/updates/F8/pending/rpy-1.0.1-4.fc8,R-2.7.0-2... https://admin.fedoraproject.org/updates/F9/pending/rpy-1.0.1-6.fc9,R-2.7.0-2...
I'm not sure why these haven't pushed yet, I did them several days ago.
~spot
r-devel@lists.fedoraproject.org
-
Alex Lancaster -
José Abílio Matos -
Patrice Dumas -
Pierre-Yves Chibon -
Tom Callaway