R devel - Oct 2007 - R can't source() long lines (PR#10383)

Full_Name: Martin Kober
Version: 2.6.0
OS: Vista & Linux
Submission from: (NULL) (137.208.185.169)


Hi!

I just stumbled upon a problem with file source()ing:

R will fail to source a file if it contains lines longer than about 8192 bytes
("input buffer overflow").

While it's save to say that human-written code won't contain lines that
long,
dump()ed data structures may reasonably contain strings longer than that (as in
my case).

What's more, R will happily help you to create such files:

x = paste(rep("12345678", 1024), collapse="")
dump("x")
source("dumpdata.R") ## this fails (on Linux & Vista)

I'm now using save/load, which is probably better in that case anyway.
Still, I
would be nice if source() worked or at least dump() would break lines to avoid
this issue.

Best regards,
Martin Kober

This is as documented in ?source, and so is not a bug.

On Tue, 30 Oct 2007, martin.kober at gmail.com wrote:
> Full_Name: Martin Kober
> Version: 2.6.0
> OS: Vista & Linux
> Submission from: (NULL) (137.208.185.169)
>
>
> Hi!
>
> I just stumbled upon a problem with file source()ing:
>
> R will fail to source a file if it contains lines longer than about 8192
bytes
> ("input buffer overflow").
>
> While it's save to say that human-written code won't contain lines
that long,
> dump()ed data structures may reasonably contain strings longer than that
(as in
> my case).
>
> What's more, R will happily help you to create such files:
>
> x = paste(rep("12345678", 1024), collapse="")
> dump("x")
> source("dumpdata.R") ## this fails (on Linux & Vista)
>
> I'm now using save/load, which is probably better in that case anyway. 
> Still, I would be nice if source() worked or at least dump() would break 
> lines to avoid this issue.
It would be even nicer if users read the documentation rather than 
expected the developers to do so for them.
> Best regards,
> Martin Kober
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
-- 
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

On 10/30/07, Prof Brian Ripley <ripley at stats.ox.ac.uk>
wrote:
> This is as documented in ?source, and so is not a bug.
Indeed. You are of course correct, sorry I missed that.

What still remains is that dump() creates a file that won't source,
without warning or stopping (since the output is basically useless).
Note that ?dump states that "A warning is issued if it is likely that
problems will arise" (in different context, but still).

Best Regards
> On Tue, 30 Oct 2007, martin.kober at gmail.com wrote:
>
> > Full_Name: Martin Kober
> > Version: 2.6.0
> > OS: Vista & Linux
> > Submission from: (NULL) (137.208.185.169)
> >
> >
> > Hi!
> >
> > I just stumbled upon a problem with file source()ing:
> >
> > R will fail to source a file if it contains lines longer than about
8192 bytes
> > ("input buffer overflow").
> >
> > While it's save to say that human-written code won't contain
lines that long,
> > dump()ed data structures may reasonably contain strings longer than
that (as in
> > my case).
> >
> > What's more, R will happily help you to create such files:
> >
> > x = paste(rep("12345678", 1024), collapse="")
> > dump("x")
> > source("dumpdata.R") ## this fails (on Linux & Vista)
> >
> > I'm now using save/load, which is probably better in that case
anyway.
> > Still, I would be nice if source() worked or at least dump() would
break
> > lines to avoid this issue.
>
> It would be even nicer if users read the documentation rather than
> expected the developers to do so for them.
>
> > Best regards,
> > Martin Kober
> >
> > ______________________________________________
> > R-devel at r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-devel
> >
>
> --
> 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
>