Hi the list Is there a way to know how many times an R package (on CRAN) has been download ? Christophe
Number of download.
6 messages · Christophe Genolini, Liviu Andronic, David Winsemius +1 more
Hello On Tue, Jan 19, 2010 at 7:51 PM, Christophe Genolini
<cgenolin at u-paris10.fr> wrote:
Is there a way to know how many times an R package (on CRAN) has been download ?
Recently this page [1] was set up, but it doesn't seem updated for some time now. Liviu [1] http://neolab.stat.ucla.edu/cranstats/
On Jan 19, 2010, at 2:51 PM, Christophe Genolini wrote:
Hi the list Is there a way to know how many times an R package (on CRAN) has been download ?
No, or at least not a comprehensive number. The question came up and was discussed in March last year. Search term "popular".: http://finzi.psych.upenn.edu/Rhelp08/2009-March/thread.html The was considerable disagreement about the validity of any such number. This is Dirk's offering, which I assume is specific to Debian and appears to be the only data offered: http://qa.debian.org/popcon.php?package=r-base More generally, it was stated that the CRAN mirroring mechanism does not support data collection of this sort. (But it appears that the UCLA server may be an exception.)
David Winsemius, MD Heritage Laboratories West Hartford, CT
On Tue, Jan 19, 2010 at 8:23 PM, David Winsemius <dwinsemius at comcast.net> wrote:
On Jan 19, 2010, at 2:51 PM, Christophe Genolini wrote:
Hi the list Is there a way to know how many times an R package (on CRAN) has been download ?
No, or at least not a comprehensive number. The question came up and was discussed in March last year. Search term "popular".: http://finzi.psych.upenn.edu/Rhelp08/2009-March/thread.html ?The was considerable disagreement about the validity of any such number. ?This is Dirk's offering, which I assume is specific to Debian and appears to be the only data offered: http://qa.debian.org/popcon.php?package=r-base More generally, it was stated that the CRAN mirroring mechanism does not support data collection of this sort. (But it appears that the UCLA server may be an exception.)
In a similar vein, has anyone ever put any 'phone home' code in a package, so that authors can track usage? Something in the package startup code that pings a logging server, for example? Yes I know doing such a thing without telling the user and giving them an opt-out is evil. Barry
On 1/19/10, Barry Rowlingson <b.rowlingson at lancaster.ac.uk> wrote:
In a similar vein, has anyone ever put any 'phone home' code in a package, so that authors can track usage? Something in the package startup code that pings a logging server, for example? Yes I know doing such a thing without telling the user and giving them an opt-out is evil.
Why would this be evil? For R, for example? I've already read some objections to this on r-help, but I'm not sure I understand the reasons. As long as the 'ping' happens once, at first start, anonymously, and requires confirmation from the user, I do not see an issue with the behaviour. Regards Liviu
On Tue, Jan 19, 2010 at 9:51 PM, Liviu Andronic <landronimirc at gmail.com> wrote:
Why would this be evil? For R, for example? I've already read some objections to this on r-help, but I'm not sure I understand the reasons. As long as the 'ping' happens once, at first start, anonymously, and requires confirmation from the user, I do not see an issue with the behaviour.
I did say 'without telling the user'.
This kind of behaviour got its bad name from closed-source software
'phoning home' and raising privacy concerns since nobody could tell
what the seemingly random stream of bytes heading off your computer
consisted of.
This kind of behaviour seems acceptible:
> library(foo)
Foo library would like to report its usage to foo.com. Would you like
it to do this every time you
use it? No personal data about you or your computer is transmitted [y/N] Y
Reporting use of library(foo) at 12:34:56 7-8-2010 to foo.com
Then next time:
> library(foo)
Reporting use of library(foo) at 12:34:56 7-8-2010 to foo.com
To stop usage reporting, do foo.reporting("off")
Barry