Skip to content
Prev 16944 / 21307 Next

[Bioc-devel] lipidr: Unable to reproduce error (possibly from S4Vectors)

Hi Ahmed,

Just a heads up that Lori's workaround (reinstalling iheatmap from 
source on all the builders) should be considered a temporary one and 
that the fundamental problem remains. The fundamental problem being that 
the iheatmap's Windows and Mac binaries that are on CRAN are broken. And 
the reason they are broken is because they contain stale information in 
their method cache.

As a consequence, the lipidr error you were seeing a few days ago on our 
build report is back today on the devel report:

   https://bioconductor.org/checkResults/3.12/bioc-LATEST/lipidr/

Note that this time it's only on Windows and Mac. This is because 
yesterday we updated R to R 4.0.2 on the devel builders so the build 
machines automatically reinstalled all the deps from scratch. iheatmap 
is one of them and the build system picked up the Windows and Mac 
binaries from CRAN, which are broken :-/

We'll see the same problem on the release builds when we update R to R 
4.0.2 there too.

Bottom line: the only true fix is that the iheatmap authors/maintainers 
upload a new version of their package to CRAN. The only change they need 
to do is bump the version. Alternatively, they could reconsider 
depending on S4Vectors. Bioconductor moves fast and CRAN packages that 
depend on Bioconductor packages are inherently at the risk of seeing 
their binaries become stale.

Hope this helps,
H.
On 6/22/20 04:50, Shepherd, Lori wrote: