R devel - Apr 2005 - documentation for 'open': some clarification?

I'm been doing more and more of file text parsing inside R instead of
coping with Perl.  For that, I need open a file and read it line-by-line.
I found the documentation for 'open' isn't very clear.

Right now it has

,----[ *help(R)[open]* ]
|  'open' opens a connection.  In general functions using connections
|  will open them if they are not open, but then close them again, so
|  to leave a connection open call 'open' explicitly.
`----

It seems that one has to call 'open' to open a live connection that can
be
parsed line by line (using readLines() or scan() to read).  But open()
cannot be directly applied to a file name.

In fact, one only needs supply the 'open' argument in file() with
something other than the empty string,

,----
| > con <- file ("S200.dat")
| > isOpen (con)
| [1] FALSE
| > con <- file ("S200.dat", open = "r")
| > isOpen (con)
| [1] TRUE
`----

It is not clear to me how 'open()' is supposed to be used.  It took me a
while to figure this out and I thought it might be worthwhile to add one
sentence or two to make it more clear in the doc.

Cheers,

Michael

-- 
Na (Michael) Li, Ph.D.               
Division of Biostatistics          A443 Mayo Building, MMC 303   
School of Public Health            420 Delaware St SE            
University of Minnesota            Minneapolis, MN 55455         
Phone: (612) 626-4765              Email: nali@umn.edu                  
Fax:   (612) 626-0660              http://www.biostat.umn.edu/~nali

GPG Public Key: 
http://www.umn.edu/lookup?SET_INSTITUTION=UMNTC&UID=nali&show_pgp=1

On Thu, 14 Apr 2005, Na Li wrote:
>
> I'm been doing more and more of file text parsing inside R instead of
> coping with Perl.  For that, I need open a file and read it line-by-line.
> I found the documentation for 'open' isn't very clear.
You appears to have missed almost all the documentation.
> Right now it has
>
> ,----[ *help(R)[open]* ]
> |  'open' opens a connection.  In general functions using
connections
> |  will open them if they are not open, but then close them again, so
> |  to leave a connection open call 'open' explicitly.
> `----
>
> It seems that one has to call 'open' to open a live connection that
can be
> parsed line by line (using readLines() or scan() to read).  But open()
> cannot be directly applied to a file name.
Not surprising, as it is documented as

      open(con, ...)
      ## S3 method for class 'connection':
      open(con, open = "r", blocking = TRUE, ...)

      con: a connection.
            ^^^^^^^^^^^
> In fact, one only needs supply the 'open' argument in file() with
> something other than the empty string,
Or use 'open(con)', but you are asking about the usage of something that
you never need to use, as should have been clear from all the examples, 
e.g. those in ?readBin (which is in the SeeAlso) that get on fine without 
it.
> ,----
> | > con <- file ("S200.dat")
> | > isOpen (con)
> | [1] FALSE
> | > con <- file ("S200.dat", open = "r")
> | > isOpen (con)
> | [1] TRUE
> `----
>
> It is not clear to me how 'open()' is supposed to be used.  It took
me a
> while to figure this out
Have you figured it out?
> and I thought it might be worthwhile to add one
> sentence or two to make it more clear in the doc.
Help pages are not tutorials: there is a reference on the help page, and 
an article in R-news.  (I have raised several times the idea of a 
technical papers section with such articles, but other do not share my 
enthusiasm.)

-- 
Brian D. Ripley,                  ripley@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

Here is an example:

# read and print first 10 lines one by one
# the next two lines could be collapsed into con <- file("myfile",
r)
con <- file("myfile") 
open(con) 
for(i in 1:10) print(readLines(con, n=1))
close(con) 

Also its possible that you may not need open.  For example,
one can just read it all in at once like this:

mylines <- readLines("myfile")
# and now mylines[1] is the first line, etc.  

# or
my.numbers <- scan("myfile")

# or
my.table <- read.table("myfile")



On 4/14/05, Na Li <nali@umn.edu> wrote:
> 
> I'm been doing more and more of file text parsing inside R instead of
> coping with Perl.  For that, I need open a file and read it line-by-line.
> I found the documentation for 'open' isn't very clear.
> 
> Right now it has
> 
> ,----[ *help(R)[open]* ]
> |  'open' opens a connection.  In general functions using
connections
> |  will open them if they are not open, but then close them again, so
> |  to leave a connection open call 'open' explicitly.
> `----
> 
> It seems that one has to call 'open' to open a live connection that
can be
> parsed line by line (using readLines() or scan() to read).  But open()
> cannot be directly applied to a file name.
> 
> In fact, one only needs supply the 'open' argument in file() with
> something other than the empty string,
> 
> ,----
> | > con <- file ("S200.dat")
> | > isOpen (con)
> | [1] FALSE
> | > con <- file ("S200.dat", open = "r")
> | > isOpen (con)
> | [1] TRUE
> `----
> 
> It is not clear to me how 'open()' is supposed to be used.  It took
me a
> while to figure this out and I thought it might be worthwhile to add one
> sentence or two to make it more clear in the doc.
> 
> Cheers,
> 
> Michael
> 
> --
> Na (Michael) Li, Ph.D.
> Division of Biostatistics          A443 Mayo Building, MMC 303
> School of Public Health            420 Delaware St SE
> University of Minnesota            Minneapolis, MN 55455
> Phone: (612) 626-4765              Email: nali@umn.edu
> Fax:   (612) 626-0660              http://www.biostat.umn.edu/~nali
> 
> GPG Public Key:
> http://www.umn.edu/lookup?SET_INSTITUTION=UMNTC&UID=nali&show_pgp=1
> 
> ______________________________________________
> R-devel@stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>