>>>>> Steve Grubb <sgr...@redhat.com>
>>>>> on Fri, 21 Jul 2017 13:53:12 -0400 writes:
> Hello Martin,
> On Friday, July 21, 2017 4:21:21 AM EDT Martin Maechler wrote:
>> I have now created an account for you.
> Thanks. Is that the preferred method of transferring these patches?
in such a case, yes .. but don't ask for a full definition of "such
a case" ;-)
If the issue may be somewhat controversial and rather in the
spirit of "I don't like what R is doing here, and I think we
should change ..."
we'd prefer it be posted here, first, in any case; but you had
no such examples.
>> >> In examples like the one below, if you have R code that shows
symptoms,
>> >> it would really help in the bug report.
>> >
>> > I am hoping that we can look at the code as seasoned programmers and
say
>> > yeah, that is a bug.
>>
>> I agree in this case.
>> OTOH, it is exactly one of the case where the bug is not
>> triggerable currently:
>>
>> al <- formals(ls); length(al) <- 3
>>
>> would trigger the bug... but you get an error message ".. vector .."
>> and as I now found that is from a slightly misguided check:
>> isVectorizable() is not approriate here and should really be
>> replaced by isList().
>>
>> So .. indeed, your report will have triggered an improvement in
>> the code, which I'm about to commit.
> That's what it's all about. :-)
>> Thank you very much Steve!
>>
>> > I run the code through Coverity and have quite a lot of
>> > problems to tell you about.
>>
>> I'm not the expert on static code analysis, but as a seasoned
>> statistician (*and* from experience with other such analyses) I
>> know that you always get false positives.
> Absolutely. I weeded the report down to 15 issues to start with. There
are
> also ways to annotate the code so that checkers dismiss something it
would
> otherwise be inclined to report.
>> >> Otherwise, someone else will have to analyze the code to decide
whether
>> >> it's a bug or missing comment. That takes time, and if there are no
>> >> known symptoms, it's likely to be assigned a low priority. The sad
>> >> truth is that very few members of R Core are currently actively fixing
>> >> bugs.
>> >
>> > That's a shame. I'd be happy to give the scan to people in core so they
>> > can see what the lay of the land looks like.
>>
>> (hmm... the above does look a teeny tiny bit arrogant in my
>> eyes; but then I'm not a native English (nor "American"
>> speaker ...)
> I apologize if that is the way it came across. "That's a shame" can also
mean
> "That's unfortunate" because I was thinking that I spent some time fixing
up
> patches that might not be wanted. However, I see that you have looked at
the
> patches and I thank you for that. :-)
> The second sentence above is an honest offer. I'd be happy to send the
output
> of the report off list (in case anything sensitive is listed). In this
and the
> other patches I haven't sent, I'm just picking the low hanging fruit.
> -Steve
Ok, thank you for the offer! In general, we would prefer public
communication of such issues because it can help to spread the
volunteer work load a bit wider than only to R Core. OTOH,
yes, there are important exceptions to this rule, as we know.
Martin
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
