Skip to content

[Bioc-devel] Package name

5 messages · Laurent Gatto, Shepherd, Lori, Wolfgang Huber +1 more

Original
Laurent Gatto
# 
The Package Guidelines for Developers and Reviewers say that:

A package name should be descriptive and should not already exist as a current package (case-insensitive) in Bioconductor nor CRAN.

The sentences says current packages - does that imply that names of packages that have been archived (on CRAN) or deprecated (on Bioconductor) are available? This is likely to lead to serious confusion.

Laurent
Shepherd, Lori
# 
Good point.  I'll open an issue on the github to fix.


Lori Shepherd

Bioconductor Core Team

Roswell Park Comprehensive Cancer Center

Department of Biostatistics & Bioinformatics

Elm & Carlton Streets

Buffalo, New York 14263
Wolfgang Huber
# 
This is probably a niche concern, but  I?d find it a pity if a good package name (*) became unavailable forever, esp. if it refers to a real-world concept not owned by the authors of the original package.
Perhaps we could allow re-using a name after a grace period (say 1 or 2 years)?
To be extra safe, one could also require the first version number of the new package be much higher than the last version of the old (dead) package.

(*) One example I have in mind where we re-used the name of an extinct project is rhdf5.

Kind regards
Wolfgang
Henrik Bengtsson
# 
For CRAN packages it's easy. Packages on CRAN are eternal. They may be
archived, but they are never removed, so in a sense they're always
"currently on CRAN". Archived packages may still be installed, but only
with some efforts of the user. Some packages go in an out of "archived"
status depending how quick the maintainer fixes issues. Because of this, I
cannot really see how a CRAN package name can be "reused" by anyone else
without a formal handover agreement between old and new maintainers. Even
so, I think CRAN needs to approve on the "update" in order to unarchive it.

Personally, I'd argue the same should apply to Bioconductor packages.
Reusing package names for other purposes/different APIs is just asking for
troubles, e.g. when it comes to future scientists trying to reproduce
legacy results.

/Henrik
On Fri, Oct 22, 2021, 03:02 Wolfgang Huber <wolfgang.huber at embl.org> wrote:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

        

  
  


  


      

    

    
2 days later

    

      
      

        


  

        

      Laurent Gatto
    

    

      
      #
    

  


  

      
My specific example falls in Henrik's category.

I am working on a package that handles peptide-spectrum matches, commonly called PSMs in proteomics. I realised, to my great dismay, that there used to be a PSM package on CRAN (https://cran.r-project.org/web/packages/PSM/index.html) for non-linear mixed-effects modelling using stochastic differential equations for population stochastic modelling. As you might imagine, that name is very far fetched in my view.

I renamed my package.