[Bioc-devel] Version bumps

[Martin Morgan]

Bioc developers!

A number of packages in the 'master' branch of their git repository 
build and check successfully, but do not propagate to the 'devel' 
repository (red light at the extreme right, e.g., ABSSeq; mouse over for 
details)

   https://bioconductor.org/checkResults/3.7/bioc-LATEST/ABSSeq

This is because the version of the git repository master branch (1.32.3) 
is less than the version of the published package in the devel 
repository (1.33.0).

The Bioconductor version scheme is described at

   https://bioconductor.org/developers/how-to/version-numbering/

Suppose a package has 'devel' version 1.3.1, and release version 1.2.0. 
The middle digit in the devel version is always odd, the middle digit in 
the release version is even and one less than devel. Best practice would 
be to introduce bug fixes and new features into the master branch and 
increment the version to 1.3.2. Bug fixes would then be ported to the 
RELEASE_3_6 branch and the release branch would be incremented to 
version 1.2.1.

Here are the packages with invalid versions, and a little commentary

 > versions %>% select(package, published, repo)
# A tibble: 9 x 3
   package        published repo
   <chr>          <chr>     <chr>
1 ABSSeq         1.33.0    1.32.3

Looks like this package had bug fixes introduced into the release 
(even-numbered) branch, and these were ported to the devel branch 
without changing version number. Introduce bug fixes in devel, and port 
to release. Remember to increment the appropriate version number in 
devel, and the appropriate version number in release. The next devel 
version number is 1.33.1.

2 esetVis        1.5.0     1.3.2

Looks like this package had bug fixes introduced into the devel branch, 
but did not pull the version bump from Biocoductor after the last 
release. The next devel version number is 1.5.1.

3 Logolas        1.3.0     1.2.1
4 microbiome     1.1.10006 1.1.2

It's true that .2 > .10006, but the number after the '.' is treated as 
an integer, and 2 < 10006. The correct version number for this package 
is 1.1.10007.

5 multiClust     1.9.0     1.8.1
6 nucleR         2.11.0    2.2.0

It looks perhaps like the nucleR author is incrementing the middle 
version field, rather than the last. The package does not propagate for 
the same reason as microbiome, 2 < 11. The rule is that the 'devel' 
version have odd-numbered middle digits, so the correct version in the 
repo would be 2.11.1.

7 QuaternaryProd 1.7.0     1.6.1
8 rDGIdb         1.5.0     1.4.2
9 Rnits          1.13.0    1.11.2

Please respond to the bioc-devel mailing list with any questions; I look 
forward to fewer red lights!

Martin


This email message may contain legally privileged and/or...{{dropped:2}}