Prof Brian Ripley <[EMAIL PROTECTED]> writes:
> It is working as documented: there is no on.exit set for str, is there?
>
> 'sys.on.exit()' retrieves the expression stored for use by
> 'on.exit' in the function currently being evaluated. (Note that
> ^^^^^^^^^^^^^^^^^^^^^^^^^
> this differs from S, which returns a list of expressions for the
> current frame and its parents.)
>
> I see you have looked at PR#269, but did not notice the crucial
> difference: here the current function is str (it is evaluating its
> arguments), not soe.test.
>
> On Mon, 7 Feb 2005 [EMAIL PROTECTED] wrote:
>
> > Full_Name: Mark Bravington
> > Version: 2.0.1
> > OS: Windows XP
> > Submission from: (NULL) (140.79.22.104)
> >
> >
> > 'sys.on.exit()' doesn't seem to be working, since R1.7.1 at least:
>
> Please read the FAQ: we don't want R-bugs clogged up with `doesn't
> seem to be working' reports, and we do say so.
>
> > soe.test <- function() {
> > on.exit( cat( 'In exit code\n'))
> > str( sys.on.exit()) # should display "language..." I think
> > 12
> > }
> >
> > (A similar bug was apparently fixed for version 0.65!)
>
> One difference being that PR#269 was a bug, and this is not.
It does, however, point to a subtlety with the sys.xxx functions,
which is liable to confuse users to the point of submitting spurious
bug reports. Perhaps we should add a note to the help page (in the
vain hope that people will read it).
Notice, BTW, that this exposes a slightly anomalous handling of the
"<-" operator. AFAIK this is common to all .Primitive calls, as
opposed to .Internal and other function calls: They do not create a
new context, hence do not increase sys.nframe() and sys.whatever in
the arguments still refer to the callers frame.
> soe.test
function() {
on.exit( cat( 'In exit code\n'))
a <- sys.on.exit() ; str(a)
12
}
> soe.test()
language cat("In exit code\n")
In exit code
[1] 12
but if you replace "<-" with a corresponding call to assign(), then
you get.
> soe.test()
NULL
In exit code
[1] 12
--
O__ ---- Peter Dalgaard Blegdamsvej 3
c/ /'_ --- Dept. of Biostatistics 2200 Cph. N
(*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918
~~~~~~~~~~ - ([EMAIL PROTECTED]) FAX: (+45) 35327907
______________________________________________
R-devel@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
