[R] reading tables from url

[Prof Brian Ripley]

On Thu, 15 Nov 2007, Duncan Temple Lang wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi Chris.
>
>  Indeed, I cannot connect to that URL either.  So I did a bit of
> digging and experimentation to find out whether one needed to
> pass additional hidden options from the form or whether the problem was
> more to do with how we connect.
>
> It turns out that the script associated with NCBI leuks.cgi is being
> fussy and wants you tell it the user agent that is performing the
> request.  (Why the two behave differently is not clear after a very
> brief look, but it is probably not worth pursuing.)
>
> AFAIR, there is no way to tell R to include a UserAgent field in the
> header of the request using url(), etc. although it did come up at one
> point.

Which is presumably why download.file works: it does set a user agent (see 
option HTTPUserAgent). I don't understand why it was added to 
download.file but not url: but that is easy to rectify.

(I don't understand why the server is set up to give an internal server 
error for a missing field, though).


-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595