nickva commented on issue #1153: Flexible replicator client auth + session 
   Thank for the comments, @janl 
   In case of the config ini / cfg files. I think there won't be many more 
config settings besides the plugin list. And even there it would not be the 
full on mfa format (like in the server side auth plugins list), just a list of 
modules separated commas.
   Ini files are not perfect, but I don't hate them, they allow overrides, 
loading from a *.d directory, stuff is put into an ets table optimized for 
concurrent reads,  we have a documentation section for it etc. The source 
config for stats work well because they are relatively static (change with 
code), it's more structured data, and there are a ton of them (that would bloat 
the ini file if it were there).
   With `update_headers` initially I started with something that took the whole 
request, body and headers, however thinking about it seemed the API was too 
broad. All use cases I could think of we only update headers, so restricted it 
back down.
   The `handle_response` is more broad because various plugins might want to 
inspect error response to make decisions how to proceed, say get a new token or 
has it be been revoked and it will never be valid again. So Basically the API 
is inherently not symmetric. The more specific name  is also helpful when 
reading the the section in the code where it is called, as it explain what it 
will do, as opposed to passing in the whole request body and wondering how much 
was really modified or what it does behind the scenes.

This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:

With regards,
Apache Git Services

Reply via email to