On Tue, 2 Jul 2024, Martin Maechler wrote:
Achim Zeileis
on Tue, 2 Jul 2024 01:05:07 +0200 (CEST) writes:
> Kevin, R-Forge is still alive but there are problems with
> the build queue as you noticed. The problems occurred
> after R-Forge upgraded to R 4.4.0 - just as there was a
> switch in the maintenance team at WU Wien.
> The problem is with the Windows build/check tools which
> work fine when run manually but either fail (if there is
> compiled code in the package) or stall when run in batch
> mode.
> Jorge, the new member in the admin team at WU, has been
> going through all the scripts with Uwe (Ligges) but they
> haven't spotted the problem so far. I think that the next
> step will be to migrate the Windows build/check processes
> to the Win-Builder machine.
> I hope that they will succeed in doing so in the next
> weeks. Apologies for the inconvenience.
> Best wishes, Achim
Thanks a lot, Achim,
for the update and explanation!
I maintain quite a few packages on R-forge,
enjoying the simplicity of svn (subversion) on the way.
For automatic building and checking, some of you may be aware
that the R-OpenSci organization has hosted and sponsored the
"R-Universe" project https://ropensci.org/r-universe/,
implemented and made quite successful I think primarily by
Jeroen Ooms. A main goal (I think) has been to *be* as
universal as possible, and provide a "portal" for "all" publicly
hosted R packages and not only foster to private company-owned
platforms such as github, but rather work with more general
"organizations" such as CRAN and BioC (each via their 'Meta
CRAN'), then also R-forge and many many more, see
https://r-universe.dev/organizations/
Consequently, the R-forge packages are also all mirrored and
presented there, *and* also built and checked,
--> https://r-forge.r-universe.dev/
... sometimes (somewhat confusingly to us, in one case) checking
against non-released versions of CRAN packages (instead of the stable
CRAN version) which of course may be useful ((and I think is
similar to how R-forge works; if there are released and
unreleased version of packages around, it's not easily
determinable *which* version should be used for tests of other
packages)).
Anyway, I was glad in one case that R-universe provided a built
version of the development version of one of my packages other
users could easily install.
Martin
PS: I have not been a fan of r-universe, notably originally, as
it diverts attention away from R-project and CRAN, but then did
notice how useful it can be to the R users community, and how
well it has been maintained, hence these kudos!
+1 Thanks for pointing to R-universe, Martin, indeed an excellent service!
An additional detail about the packages at:
https://R-Forge.R-universe.dev/
This are the development versions of the R-Forge packages that are already
released on CRAN.
If anyone has R-Forge packages that are not (yet) released on CRAN and you
want to deploy them via R-universe, then you can do so by setting up your
own universe. I'm using that for many of my R-Forge packages, see
https://zeileis.R-universe.dev/
and
https://github.com/zeileis/zeileis.r-universe.dev
for the underlying configuration.
Best wishes,
Achim
> On Tue, 2 Jul 2024, Kevin R. Coombes wrote:
>> Hi,
>>
>> I have been maintaining packages in R-Forge for many
>> tears. Last week I sent an email to r-fo...@r-project.org
>> to report problems with the build process. It appears
>> that any changes I have pushed to R-Forge over
>> approximately the last two months have resulted in the
>> package remaining in the "Building" state, even though
>> the logs suggest that the package built successfully on
>> both LINUX and Windows. (Also, four of the six affected
>> packages only included changes to the man pages to clean
>> up NOTEs from the R cmd checks on old versions at CRAN,
>> where the new versions now happily reside.) I have
>> received no response nor acknowledgement to my email to
>> R-Forge.
>>
>> Assuming that R-Forge has finally succumbed to the
>> ravages of entropy, does anyone have advice on creating a
>> git project that contains multiple R packages? (I really
>> don't want to have to create 20+ new git projects, one
>> per package).
>>
>> Best, Kevin
>>
>> ______________________________________________
>> R-package-devel@r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-package-devel
>>
> ______________________________________________
> R-package-devel@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel
______________________________________________
R-package-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel
