Full_Name: Rick Sayre
Version: 2.9.0
OS: windows, os x, linux
Submission from: (NULL) (138.72.153.236)
if a tcl "string" contains binary data, there is a problem getting this data
to R, since there appears to be no way to get "raw" data without NULL
interpretation. Here is a simple example. The data should be
Here are 1<one NULL>Here is 2<two NULLs>And finally three<three NULLS>
> tq <- .Tcl("set tcls [binary format H*
\"48657265206172652031004865726520697320320000416E642066696E616C6C79207468726565000000\"]")
#tclvalue stops returning data after the first NULL
> tclvalue("tcls")
[1] "Here are 1"
#interestingly, as.character gets more, but gets confused
> as.character(tq)
[1] "Here" "are" "1" "is" "2" "finally" "three"
> # we confirm that the tcl representation does indeed have the NULLs:
> .Tcl("binary scan $tcls H* tmps")
<Tcl> 1
> tclvalue("tmps")
[1]
"48657265206172652031004865726520697320320000416e642066696e616c6c79207468726565000000"
I would naively suggest something like tclrawvalue() and/or as.raw() utilizing
Tcl_GetByteArrayFromObj
A good discussion of the basics of this issue for general tcl embedding can be
found here:
http://www2.tcl.tk/1180
For those who might be wondering "what the heck are you doing trying to pass
binary data from tcl?", the answer is that non-blocking I/O does not work on
windows,
and timeouts, modes, and line control for RS232 ports aren't supported at all on
R.
Using tcl provides a nice portable way to do this in R, but currently requires
converting everything to ascii encodings since the binary data fails to make it
across.
Cheers
--Rick
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
