[Bioc-devel] Build error "failure: length > 1 in coercion to logical" not reproducible

[bode@hoier m@iii@g oii bioi@i@jku@@t]

Thanks a lot for your swift and helpful reply, Henrik! Yes, I could reproduce the error on my local system by adding "_R_CHECK_LENGTH_1_LOGIC2_=verbose" to my .Renviron file. I will fix the errors and push fixes soon.

Thanks again,
Ulrich


-----Original Message-----
From: Henrik Bengtsson <henrik.bengtsson using gmail.com> 
Sent: Donnerstag, 17. Juni 2021 17:21
To: bodenhofer using bioinf.jku.at
Cc: bioc-devel using r-project.org
Subject: Re: [Bioc-devel] Build error "failure: length > 1 in coercion to logical" not reproducible

On Thu, Jun 17, 2021 at 2:32 AM <bodenhofer using bioinf.jku.at> wrote:
>
> Dear colleagues,
>
> It seems to me that, starting with the latest BioC devel branch 
> (3.14), the build systems have become more pedantic about logical 
> vectors of length > 1 in conditions. Two of the packages I am maintaining, 'kebabs' and 'procoil'
> currently fail to build. Surely I want to fix this. However, I cannot 
> reproduce these errors on my local system (R 4.1.0 alpha on Ubuntu 
> 18.04 LTS). The discussions 
> https://support.bioconductor.org/p/9137605/ and
> https://github.com/Bioconductor/BBS/issues/71  have pointed me to the 
> setting "_R_CHECK_LENGTH_1_CONDITION_=verbose".
>
> First question: Can anybody confirm that this has been changed in the 
> recent devel?

Not a Bioc maintainer, but yes, the Bioc build system added this on May 22, 2021 in order to catch similar bugs in package vignettes, cf.
https://community-bioc.slack.com/archives/CLUJWDQF4/p1622062783020300?thread_ts=1622053611.008100&cid=CLUJWDQF4

>
> Second question: I have tried to include 
> "_R_CHECK_LENGTH_1_CONDITION_=verbose" in my .Renviron file and it 
> seems that my R session respects that. However, when I run 'R CMD 
> build' on the aforementioned packages, they still build fine. The 
> suggestions in
> https://github.com/Bioconductor/BBS/issues/71 don't work for me either 
> (maybe I have done something wrong). I would actually like to 
> reproduce the errors in my local system, since this will help me 
> fixing the errors and testing the changes. So can anybody give me 
> advice how I can make my local installation to check for logical 
> vectors of length > 1 in conditions more strictly?

You want to set:

_R_CHECK_LENGTH_1_LOGIC2_=verbose

That one catches bugs where x && y or x || y is called with length(x)
> 1 or length(y) > 1.

Using:

_R_CHECK_LENGTH_1_CONDITION_=verbose

catches bugs where if (x) { ... } and similar conditions are called with length(x) > 1.

In your case, a reproducible minimal example is:

Sys.setenv("_R_CHECK_LENGTH_1_LOGIC2_"="verbose")
files <- c("a", "b")
files <- Rsubread:::.check_and_NormPath(files)
...
Error in is.na(files) || is.null(files) :

  'length(x) = 2 > 1' in coercion to 'logical(1)'

The problem is that there's a is.na(files) || is.null(files) in the code, where

> is.na(files)
[1] FALSE FALSE
> is.null(files)
[1] FALSE

so, we have an x || y case with length(x) > 1.

/Henrik




>
> Thanks a lot in advance,
> Ulrich
>
> _______________________________________________
> Bioc-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel