Skip to content

[Bioc-devel] LazyData in DESCRIPTION file

5 messages · Vinh Tran, Vincent Carey, Kasper Daniel Hansen

Original
Vinh Tran
# 
Dear Bioc members,

I have just encountered a warning during the CHECK that some data objects are used in the documents but not in code (e.g. ?Variables with usage in documentation object ?ppTree? but not in code"). They are the demo data, that I am using only in the examples for demonstrate the usage of some functions. Adding LazyData: True to the DESCRIPTION can solve that issue, but according to the package guidelines it is not recommended. Could you please show me what should I do in this case? The demo data is only about 15 KB at max. 

Many thanks for your advices!

Best regards,
Vinh

--------------------------------
Dr. Vinh Tran

Dept. for Applied Bioinformatics
Inst. for Cell Biology and Neuroscience
Goethe University Frankfurt

Biologicum, Room 3.209
Phone +49 (0)69/798-42118
Vincent Carey
# 
I see this is guideline 7 at
https://bioconductor.org/developers/package-guidelines/

I have used LazyData: TRUE so that [pkgname]::[entity] can be used instead
of data().  The
claim that it is "rarely a good thing" and slows down package loading can
be weighed against
convenience.  I am not sure you should use LazyData to avoid a
documentation warning
however.  Can you give more details on what package is generating the
warning?
On Wed, Apr 29, 2020 at 5:34 PM Vinh Tran <tran at bio.uni-frankfurt.de> wrote:

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Kasper Daniel Hansen
    

    

      
      #
    

  


  

      
If you don't have lazy data on, you should be able to load the data by
using data().  Let us say you example data is called expData.

With lazy data
  print(expData)

Without lazy data you need explicit loading
  data(expData)
  print(expData)

For the purpose of examples, there is also a user-level advantage of using
an explicit data() statement: the user can see that the data comes from
somewhere, instead of appearing out of nowhere. For examples, I actually
only see advantages of not using lazyData, because of these explicit
statements.

My historical impression: I would say that when lazyData was introduced, it
seems to me that the intention was widespread use. It seems to me that the
tides have turned against lazy data and the official recommendation is to
not use it unless you have good reasons.  One disadvantage with widespread
use of lazyData is that the names of these objects have to be accessible
somewhere.

Note: one thing I have realized very belatedly is that the lazyData field
is a boolean, the right statements are one of
  lazyData: TRUE
  lazyData: FALSE
For example, I think it has to be all uppercase.

Best,
Kasper

On Wed, Apr 29, 2020 at 6:05 PM Vincent Carey <stvjc at channing.harvard.edu>
wrote:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Vinh Tran
    

    

      
      #
    

  


  

      
Dear Kasper,

Many thanks for your reply. Yes, I am using data() to load the data. However, the problem is, when I checked my package using CMD Check, it threw me the warning that I am using that data only in the example but not in code. I see the disadvantages of LazyData, but I just want to know in this case, how can I solve that issue by not using LazyData. If you have any solutions, I am very appreciated! 

Btw, I added ?LazyData: true? to the DESCRIPTION and the warning went away. So, I am not sure if it must always be uppercase.

Best regards,
Vinh

--------------------------------
Dr. Vinh Tran

Dept. for Applied Bioinformatics
Inst. for Cell Biology and Neuroscience
Goethe University Frankfurt

Biologicum, Room 3.209
Phone +49 (0)69/798-42118

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Kasper Daniel Hansen
    

    

      
      #
    

  


  

      
This seems really weird, and I don't think you should be using lazyData to
get a message to disappear.  Is there a GitHub page where we can browse the
code and do you have the full output of R CMD check somewhere?

        
On Thu, Apr 30, 2020 at 9:05 AM Vinh Tran <tran at bio.uni-frankfurt.de> wrote: