[RkWard-devel] SPSS test files

[Thomas Friedrichsmeier]

On Sunday 04 February 2007 16:31, you wrote:
> So if it is as I think, I do not know if there is a way for that to be
> fixed even from R-developers, and if it can be fixed I do not know if they
> want to. Does it make sense to ask something like this or reported as a
> bug? From the answers to Thomas question on r-help list I believe that they
> do not think this as bug. It is just the way it works.

Well, the R folks are pretty reluctant at viewing something as a bug. While 
this is sometimes not nice, and the tone on the mailing lists is a bit rough, 
sometimes, it's probably a good thing. It means you can be fairly sure that 
there are no dramatic changes unless needed, so old code generally keeps 
working.

In this particular case, at least Thomas Lumley (who seems to be the one who 
feels most responsible for the read.spss code) wrote that it would be nice, 
if read.spss did automatic encoding conversion (but said it's not a priority 
to him).

> > read.spss("~/Desktop/12.sav")
>
> Error in read.spss("~/Desktop/12.sav") : error reading system-file header
> In addition: Warning message:
> ~/Desktop/12.sav: position 0: Variable name begins with invalid character

This one, I think you should still report (first to r-help, with a test file, 
and full example code). After all, it really is a total failure to read that 
file, and nothing that can be worked around. I don't know, whether they will 
fix this too soon (I did have a look at the do_read_spss C code, and the PSPP 
code that could be used as an update, and I can see that it's not exactly 
easy to fix this), but they should be made aware of this.

> sav12 comes 2.sav and I cannot see it
>
> > sav12
>
> Error: invalid multibyte string

This one at least (failure to print these strings), Peter Dalgard has 
indicated will be fixed in R 2.5.

It probably should be possible to convert these strings to the current 
encoding, by using the rk.temp.convert function I use in the SPSS import 
plugin, right? Probably we should also have a stand-alone "Convert encoding" 
plugin, but that's something for later.

Regards
Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/rkward-devel/attachments/20070204/e26d0e2b/attachment.sig>