Shazwazza opened a new issue #396:
URL: https://github.com/apache/lucenenet/issues/396


   Decided to create a central discussion about this here instead of all of the 
ideas getting lost in various threads :)
   
   Here's some found previous notes
   
   - https://github.com/apache/lucenenet/issues/282#issuecomment-635345730 
(note: a lot has changed from this discussion to now!)
   - https://github.com/apache/lucenenet/pull/393
   - https://github.com/apache/lucenenet/pull/231
   - https://github.com/apache/lucenenet/pull/229#issuecomment-520831453 
(scroll down to the "code samples" heading)
   
   The issue is: We have a [conversion 
program](https://lucenenet.apache.org/contributing/documentation.html#api-docs) 
that converts the Lucene (java) docs to a doc format that we can use for 
Lucene.Net. Each time this is run it will overwrite our docs files. This means 
that any manual updates made to these docs files such as fixing spelling 
mistakes and more importantly code snippets would just be overwritten. 
   
   ## I have a simple proposal
   
   In various old threads we had some complex proposals but none of these are 
'perfect' and won't cater very well for things like spelling mistakes. Instead 
of inventing some system, we can simply use Git:
   
   * We have a branch for each lucene (java) docs conversion that we plan to do
     * currently this is just one: 4.8.1. This branch can be called something 
like `docs/converted/4.8.1` and created from `master`
   * These `converted` branches are the only branches we ever execute the 
`JavaDocToMarkdownConverter` conversion program. 
     * So currently, we would run `JavaDocToMarkdownConverter` once we create 
this branch. There will probably be no changes now since we've already executed 
this in master.
   * Merge the `converted` branch to the master branch
   * Make any required changes to the master branch docs files
   * If we need to re-run `JavaDocToMarkdownConverter` because we've made some 
fixes to it to fix some conversions, formatting, etc... we re-run this on the 
`converted` branch, then merge changes to the master branch. This will trigger 
a bunch of merge conflicts which can be resolved by Git merge. 
   
   We won't be running this conversion program that often so the amount of 
merging would be minimal and we'd only have merge conflicts on files that we've 
changed that have large conflicts with changes we've made in the converter ... 
which would be quite rare so I don't foresee a lot of work with a Git merge.
   
   Some basic rules to this:
   
   * We will never merge into a `converted` branch from master since this will 
mean merge conflicts are not triggered and everythig would just be overwritten 
again
   * When we start working on a newer version of Lucene.Net, a new `converted` 
branch will be created from the current `converted` branch and the conversion 
is run there for the new lucene version. 
   
   @NightOwl888  Any thoughts on that?
   
   


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to