An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20110411/41264f4e/attachment.pl>
CRAN problem with plyr-1.4.1
9 messages · Douglas Bates, Ian Davis, Hadley Wickham +1 more
The first thing to do is try another mirror. The "official" (or as official as we ever get about anything) U.S. mirror is http://cran.us.R-project.org They tend to be very good about updating. Presently the source package for plyr is at version 1.5 and the binary versions are both at 1.4.1
On Mon, Apr 11, 2011 at 1:27 PM, Ian Davis <ian.w.davis at gmail.com> wrote:
It looks like there might be some kind of problem with the Plyr-1.4.1 packages pushed to CRAN? ?The web pages show 1.4.1 as the current version, but trying to fetch the source through the provided link gives a 404: http://lib.stat.cmu.edu/R/CRAN/web/packages/plyr/index.html $ wget http://lib.stat.cmu.edu/R/CRAN/src/contrib/plyr_1.4.1.tar.gz --2011-04-11 13:19:09-- http://lib.stat.cmu.edu/R/CRAN/src/contrib/plyr_1.4.1.tar.gz Resolving lib.stat.cmu.edu... 128.2.241.212 Connecting to lib.stat.cmu.edu|128.2.241.212|:80... connected. HTTP request sent, awaiting response... 404 Not Found 2011-04-11 13:19:09 ERROR 404: Not Found. This prevented me from installing ggplot2 until I went back and found an old version (1.4) of Plyr to install manually. ?Since it looks like Plyr was *just* updated a few days ago, I'm guessing something went awry? ?I checked several CRAN mirrors and got the same problem with all of them. ?They think the current version is 1.4.1, but they don't have any files available for download. Hope this helps, Ian ? ? ? ?[[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
On 11.04.2011 21:03, Douglas Bates wrote:
The first thing to do is try another mirror. The "official" (or as official as we ever get about anything) U.S. mirror is http://cran.us.R-project.org They tend to be very good about updating. Presently the source package for plyr is at version 1.5 and the binary versions are both at 1.4.1
Right, and on CRAN master the update process was running and hence you got a 440. Or in other words: Please just be a bit patient after a package was updated, the automated processes may take a day to complete. You can access the repositories without any problem using install.packages() and friends. Uwe Ligges
On Mon, Apr 11, 2011 at 1:27 PM, Ian Davis<ian.w.davis at gmail.com> wrote: It looks like there might be some kind of problem with the Plyr-1.4.1 packages pushed to CRAN? The web pages show 1.4.1 as the current version, but trying to fetch the source through the provided link gives a 404: http://lib.stat.cmu.edu/R/CRAN/web/packages/plyr/index.html $ wget http://lib.stat.cmu.edu/R/CRAN/src/contrib/plyr_1.4.1.tar.gz --2011-04-11 13:19:09-- http://lib.stat.cmu.edu/R/CRAN/src/contrib/plyr_1.4.1.tar.gz Resolving lib.stat.cmu.edu... 128.2.241.212 Connecting to lib.stat.cmu.edu|128.2.241.212|:80... connected. HTTP request sent, awaiting response... 404 Not Found 2011-04-11 13:19:09 ERROR 404: Not Found. This prevented me from installing ggplot2 until I went back and found an old version (1.4) of Plyr to install manually. Since it looks like Plyr was *just* updated a few days ago, I'm guessing something went awry? I checked several CRAN mirrors and got the same problem with all of them. They think the current version is 1.4.1, but they don't have any files available for download. Hope this helps, Ian [[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20110412/d9300129/attachment.pl>
On 12.04.2011 14:52, Ian Davis wrote:
Uwe, I really do appreciate the help from you and Douglas on resolving this problem. However, I respectfully disagree on a few points: - install.packages() was broken and failing to install successfully.
install.packages() works well, it is *not* broken.
This is why I was tracking down the URL via the CRAN website.
yes, but that was not recent since the repository was more recent. And updating web pages takes its time for 3000 packages. download.packages() would have done it fine, since it is not relying on the web page. But then, install.packages() worked as well.
- I think it's crazy to have an update process that breaks links for ~24 hours every time a package is updated.
No, I just said wait for 24 hours to be on the safe side. I does not take 24 hours. For binary package versions, you need to wait 72 hours (at least that is what we promise). > This means that on any given
day, some significant fraction of R packages just cannot be installed?
They can always be installed, since a repository is up to date at once, just the web pages are not updates at once.
Even if you "just" have broken links on the web site, it looks pretty bad. "Wait till tomorrow" seems like a poor response for a project as large and widely used as R. I'll repeat what I told Douglas -- as a user of R, it would be very helpful if the list of mirrors indicated which one(s) were authoritative (and by implication, which are unreliable).
Unreliable mirrors are removed from the list of mirrors from time to time. Mirrors are updates at least twice a week. Best, Uwe Ligges
Thanks again for your help, and for your work on maintaining R.
Best,
Ian
2011/4/12 Uwe Ligges <ligges at statistik.tu-dortmund.de
<mailto:ligges at statistik.tu-dortmund.de>>
On 11.04.2011 21:03, Douglas Bates wrote:
The first thing to do is try another mirror. The "official" (or as
official as we ever get about anything) U.S. mirror is
http://cran.us.R-project.org
They tend to be very good about updating. Presently the source
package for plyr is at version 1.5 and the binary versions are
both at
1.4.1
Right, and on CRAN master the update process was running and hence
you got a 440. Or in other words: Please just be a bit patient after
a package was updated, the automated processes may take a day to
complete. You can access the repositories without any problem using
install.packages() and friends.
Uwe Ligges
On Mon, Apr 11, 2011 at 1:27 PM, Ian
Davis<ian.w.davis at gmail.com <mailto:ian.w.davis at gmail.com>>
wrote:
It looks like there might be some kind of problem with the
Plyr-1.4.1
packages pushed to CRAN? The web pages show 1.4.1 as the
current version,
but trying to fetch the source through the provided link
gives a 404:
http://lib.stat.cmu.edu/R/CRAN/web/packages/plyr/index.html
$ wget
http://lib.stat.cmu.edu/R/CRAN/src/contrib/plyr_1.4.1.tar.gz
--2011-04-11 13:19:09--
http://lib.stat.cmu.edu/R/CRAN/src/contrib/plyr_1.4.1.tar.gz
Resolving lib.stat.cmu.edu... 128.2.241.212
Connecting to lib.stat.cmu.edu
<http://lib.stat.cmu.edu>|128.2.241.212|:80... connected.
HTTP request sent, awaiting response... 404 Not Found
2011-04-11 13:19:09 ERROR 404: Not Found.
This prevented me from installing ggplot2 until I went back
and found an old
version (1.4) of Plyr to install manually. Since it looks
like Plyr was
*just* updated a few days ago, I'm guessing something went
awry? I checked
several CRAN mirrors and got the same problem with all of
them. They think
the current version is 1.4.1, but they don't have any files
available for
download.
Hope this helps,
Ian
[[alternative HTML version deleted]]
______________________________________________
R-help at r-project.org <mailto:R-help at r-project.org> mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible
code.
______________________________________________
R-help at r-project.org <mailto:R-help at r-project.org> mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20110412/c8cd8e96/attachment.pl>
On 12.04.2011 17:08, Ian Davis wrote:
2011/4/12 Uwe Ligges <ligges at statistik.tu-dortmund.de
<mailto:ligges at statistik.tu-dortmund.de>>
On 12.04.2011 14:52, Ian Davis wrote:
I really do appreciate the help from you and Douglas on
resolving this
problem. However, I respectfully disagree on a few points:
- install.packages() was broken and failing to install successfully.
install.packages() works well, it is *not* broken.
Perhaps I was unclear in my choice of words -- install.packages() *is*
not "broken" in the general sense; it does work well, most of the
time. However, at the time I reported the problem, it *was* failing to
install plyr, using three different mirrors. That is,
install.packages("plyr")
gave an error message and did not successfully install the package,
whether I specified the mirror at CMU, NIH, or UNC-CH. That's all I
meant by "broken". I understand that you believe this *should not* or
even *cannot* happen, but I'm telling you that it did. This may
indicate there is a bug in the mirroring process, in that, under
exceptional conditions yet to be identified, install.packages() et al.
are unable to retrieve correct, consistent information from a mirror for
a period of time.
Then, can we have the ERROR message, please? Otherwise the only explanation I can guess is that a mirror grabs the contents of a repository exactly in the second the repository is updated and that is unlikely, particularly if more than one mirror is involved.
Again, to be clear, I'm not protesting that the *newest* packages were unavailable on these three mirrors. I'm protesting that *no version at all* of plyr was available from these three mirrors, using install.packages(), at the time I reported the problem. I'll leave it at this -- I'm only trying to help out by reporting a bug in the process. If no one cares, I'll shut up and sit down.
We do care, but the mirrors seem to be fine and hence we cannot debug what you have seen before. Best, Uwe
Regards, Ian
Then, can we have the ERROR message, please? Otherwise the only explanation I can guess is that a mirror grabs the contents of a repository exactly in the second the repository is updated and that is unlikely, particularly if more than one mirror is involved.
Isn't one possible explanation that PACKAGES.gz on the mirror was updated before the package directory was? That would seem to me a plausible hypothesis to me, because rsync seems to send files in the top level directory before files in directories below. Hadley
Assistant Professor / Dobelman Family Junior Chair Department of Statistics / Rice University http://had.co.nz/
1 day later
On 12.04.2011 17:29, Hadley Wickham wrote:
Then, can we have the ERROR message, please? Otherwise the only explanation I can guess is that a mirror grabs the contents of a repository exactly in the second the repository is updated and that is unlikely, particularly if more than one mirror is involved.
Isn't one possible explanation that PACKAGES.gz on the mirror was updated before the package directory was?
It is in the same directory. Uwe
That would seem to me a plausible hypothesis to me, because rsync seems to send files in the top level directory before files in directories below. Hadley