R devel - Mar 2013 - Printing warning messages around R_tryEval

Hi!

In RKWard we use R_tryEval() at a number places to run R commands. In order to 
make sure that any generated warnings become visible close to the problem, we 
were following up (most of) these calls with Rf_PrintWarnings().

Rf_PrintWarnings() was never available in the headers (as far as I know), and 
in r61771 the symbol has been hidden from libR.so. So this solution is no 
longer available. Could you give advice on the best way to print warnings 
before returning to the top level?

Some available options, I am aware of:
1. Call R function warnings(). However, this will always print the latest 
warning, even if it was generate by some earlier command. I would need a way 
to print new warnings, only.

2. Use options(warn=1) where applicable. However, in some cases, collecting 
warnings until some procedure is complete, and printing them then, would be 
preferrable.

3. I see there is an internal call printDeferredWarnings(), which seems to be 
almost exactly what I want. However, using an .Internal() does not look like a 
terribly stable solution, either. Also, having direct access to a similar 
function from the C-API would be very comfortable.

Thanks!
Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL:
<https://stat.ethz.ch/pipermail/r-devel/attachments/20130306/af4f758a/attachment.bin>

We are having a similar issue with rpy2 and R-devel.
I would also vote for having back a C-level solution to the problem.

I cannot find an explicit explanation for the change in the SVN logs,
and traced the change to rev 61771:

------------------------------------------------------------------------
r61771 | ripley | 2013-01-29 12:09:45 +0100 (Tue, 29 Jan 2013) | 1 line

more hiding
------------------------------------------------------------------------

Could this mean that the change could reverted and Rf_PrintWarnings
just be added to the C API ?

Laurent


On 2013-03-07 12:00, r-devel-request at r-project.org
wrote:
>
> Hi!
>
> In RKWard we use R_tryEval() at a number places to run R commands. In order
to
> make sure that any generated warnings become visible close to the problem,
we
> were following up (most of) these calls with Rf_PrintWarnings().
>
> Rf_PrintWarnings() was never available in the headers (as far as I know),
and
> in r61771 the symbol has been hidden from libR.so. So this solution is no
> longer available. Could you give advice on the best way to print warnings
> before returning to the top level?
>
> Some available options, I am aware of:
> 1. Call R function warnings(). However, this will always print the latest
> warning, even if it was generate by some earlier command. I would need a
way
> to print new warnings, only.
>
> 2. Use options(warn=1) where applicable. However, in some cases, collecting
> warnings until some procedure is complete, and printing them then, would be
> preferrable.
>
> 3. I see there is an internal call printDeferredWarnings(), which seems to
be
> almost exactly what I want. However, using an .Internal() does not look
like a
> terribly stable solution, either. Also, having direct access to a similar
> function from the C-API would be very comfortable.
>
> Thanks!
> Thomas