Skip to content

Upgrading R

5 messages · Gorjanc Gregor, Dirk Eddelbuettel, Uwe Ligges +2 more

#
Hello!

New version of R has came out and I would like to thank to all developers
on this matter. So I should probably upgrade. Fine and no problem. For 
windows I just grab the latest precompiled binnaries and install them. Then
I see a report on a bug, which is or will be fixed in pacthed version. So
I need to get binnaries from patched build and install them, right?

How often do you people upgrade R on windows? For every patch? I know that
it depends on the bug, but I would just like to hear what are your habits.

I suppose Debian packages of "base R" are updated accordingly to R patches,
aren't they?

Thanks in advance!

--
Lep pozdrav / With regards,
    Gregor Gorjanc

------------------------------------------------------------------------
University of Ljubljana
Biotechnical Faculty       URI: http://www.bfro.uni-lj.si/MR/ggorjan
Zootechnical Department    email: gregor.gorjanc <at> bfro.uni-lj.si
Groblje 3                  tel: +386 (0)1 72 17 861
SI-1230 Domzale            fax: +386 (0)1 72 17 888
Slovenia
#
On 24 April 2005 at 14:45, Gorjanc Gregor wrote:
| I suppose Debian packages of "base R" are updated accordingly to R patches,
| aren't they?

No, I tend to follow R Core and make release when actual minor release are
made. On the other hand, I try to help with alpha and beta releases during
the build-up to a release.

Full details are of course in the changelog, on your Debian box in
/usr/share/doc/r-base-core/changelog.Debian.gz and on the web at
http://changelog.debian.net/r-base

Regards, Dirk
#
Gorjanc Gregor wrote:

            
The binaries on CRAN are updated by Duncan Murdoch frequently. Of 
course, you can get the sources at arbitrary time and compile yourself 
hourly, if you prefer. ;-)

You have to distinguish official releases and patched versions.

For official releases:
Form the developer page: "The general schedule is to have major releases 
(x.y.0) biannually".
During the last few years, each major release was followed by one minor 
(bug-fix) release after roughly one month.

Uwe Ligges
#
Uwe Ligges wrote:
At present, "frequently = daily" is the target, but occasionally I miss 
the target for various reasons, such as a build error or something wrong 
with my job scheduling.  Usually I notice within a few days and get 
things going again.
Right, the daily builds are unreleased snapshots, and are not tested by 
me.  If all goes well then whoever committed the latest changes tested 
them and didn't break anything, but things occasionally go wrong, which 
is why we have the alpha and beta test periods before a release.

Duncan Murdoch
#
Duncan Murdoch <murdoch at stats.uwo.ca> writes:
In particular, daily snapshots are usually only tested on the machine
of the last person to commit code to the repository. One of the main
points of having the test periods before release, and the associated
code freezes, is that this allows a reasonable chance that the released
sources work across a range of platforms. (The other reason is that it
prevents developers from rushing in changes.)