segfault in browseURL()

Using Win2K (and reportedly WinXP), when the length of the 'url' string

= 280 characters, a segmentation fault occurs.

This doesn't seem to be affecting unix machines.

With what settings of argument `browser'?

If you absolutely must have such long URLs, try specifying a browser 
(although you will probably find limits that way too).

As the FAQ asks, we do need a reproducible example to check the fixes.

With what settings of argument `browser'?

In this case, none.

getOption("browser") returns NULL.  When browseURL() is working form e
under that configuration it pulls up IE5

If you absolutely must have such long URLs, try specifying a browser 
(although you will probably find limits that way too).

When I tried specifying 'browser="explorer"', I still get errors, although
not segfaults.  The error seems to be dependent on exactly what the URL
string is, and is always fairly odd (as if Windows is pushing the extra
bits off into other commands).

As the FAQ asks, we do need a reproducible example to check the fixes.

Well, the toy example I was using to first verify that it was coming from
browseURL in general was just to do this:
z <- rep("z", 300)
z <- paste(z, collapse="")
browseURL(z)

While that URL obviously won't work, note that if you make it of a smaller
length that instead of a segfault you should just get the error that the
URL doesn't seem to exist.

The other example I was using (a 'real' example) requires the use of the
annotate and hgu95av2 packages from Bioconductor (and was supplied by
James MacDonald):

library(annotate)
data(eset)
gn <- geneNames(eset)[453]
gn <- getPMID(gn,"hgu95av2")
gn <- unlist(gn, use.names=FALSE)
pubmed(gn, disp="browser")

This builds up a URL query and then calls 'browseURL(query)'.

Well, the toy example I was using to first verify that it was coming from
browseURL in general was just to do this:
z <- rep("z", 300)
z <- paste(z, collapse="")
browseURL(z)

That's not a URL at all, and I get nothing (as I should). If I put http:// 
in front it works (as a search item).

This builds up a URL query and then calls 'browseURL(query)'.

And you could extract `query' and tell us what that is ....

Well, the toy example I was using to first verify that it was coming from
browseURL in general was just to do this:
z <- rep("z", 300)
z <- paste(z, collapse="")
browseURL(z)

That's not a URL at all, and I get nothing (as I should). If I put http:// 
in front it works (as a search item).

That's not the point.  When it is repped to length 300, it causes a

segfault for me.  When it is repped to a length of say, 250 - it simply
doesn't work properly (as one would expect, because as you so correctly
pointed out 'aaaaaaa....' isn't a URL.  My point here was to demonstrate
the segfaulting due to excessively long strings, which at least for me,
does not seem to be tied to a URL being valid or not.

Here:

z <- paste("http://www.r-project.org/", paste(rep("a", 200),
           collapse=""))
browseURL(z)

This gives an error that the URL does not exist.

z <- paste("http://www.r-project.org/", paste(rep("a", 300),
           collapse=""))
browseURL(z)

This causes a segfault.

This builds up a URL query and then calls 'browseURL(query)'.

And you could extract `query' and tell us what that is ....

[1]
"http://www.ncbi.nih.gov/entrez/query.fcgi?tool=bioconductor&cmd=Retrieve&db=PubMed&list_uids=12730033%2c12691826%2c12544996%2c12490434%2c12477932%2c12411538%2c12391142%2c12207910%2c11971973%2c11864979%2c10859165%2c10216320%2c10205060%2c3931075%2c3470951%2c3019832%2c2880793%2c2858050%2c2538825%2c1700760%2c1478667"