thisisnic edited a comment on issue #10:
URL: https://github.com/apache/arrow-cookbook/issues/10#issuecomment-894213354


   > First, it would probably be good if each implementation had the freedom to 
adopt a strategy that works for them...
   
   Agreed
   
   Thinking specifically about the R cookbook then: 
   
   > 1...
   
   I don't think this would be suitable for the R version of the cookbook, as 
the API is growing in features at the moment, and it may be the case that users 
work for organisations which aren't going to let them keep up with the 3 
monthly release cycle - this could be confusing when readers are faced with 
content which doesn't work for them.
   
   > 2...
   
   I thought about this - it would make reordering content complicated, and I 
don't know if that's an issue or not.  I don't think it'd be too hard to 
implement though.
   
   > 3...
   
   This approach sounds closer to what I was thinking, but could be tricky to 
implement in terms of integrating it with `bookdown`.
   
   > The one approach I think we should NOT do...
   
   Yeah, agreed, I think we need to be able to have something for the current 
version (nightlies) 
   
   How about this approach:
   
   4.  a modified version of 3:
   
   * instead of a dropdown, multiple versions of the cookbook are built, and so 
the URL for the main page is e.g. `whatever/arrow-cookbook/r/5.0.0/index.html`. 
 Therefore, the differing functionality is kept separate to prevent confusing 
readers.
   
   * the content relating to the upcoming release is merged into `main`, which 
builds against the nightlies, and once a release has been done, that version of 
the cookbook gets its own branch. 
   
   * admittedly, there is a bit of extra work when a new PR is merged, as it 
would be necessary to create additional PRs to add the content to previous 
versions if the recipe works on those.  I suppose this extra workload might be 
higher at first, but as time goes on and more recipes are added, new content is 
more likely to be applicable only to new features rather than existing ones
   
   * with each release, we add a new CI job for that release.  I suppose a 
potential issue is this getting potentially huge, but we could choose to adopt 
a strategy like "freezing" older cookbooks so that new recipes are never added 
to them and so they don't need to be rebuilt, and only building against the 
most recent patch for each major release.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


Reply via email to