potiuk edited a comment on issue #19250:
URL: https://github.com/apache/airflow/issues/19250#issuecomment-954080195


   > In essence, this approach does not change the strategy of using DAG, but 
it can give some users who do not have git repository another choice to deploy 
on the cloud.
   
   This is all fine and I agree with giving people more options. I do not argue 
with that (actually - I do not even argue at all- i ask others in the community 
for opinion and present my own).
   
   However there is a huge difference how the options for the user are provided;
   
   1) if it is provided as ecosystem/external - the person/organisation who 
provides it provides all the support, user questions, solving problems. 
customisations
   
   2) if it is provided by the Apache Airflow Communtiy - it is the community 
that provides the support, release working software, fixes problems and 
generally supports the users.
   
   Both are viable options. Both has happened in the past  for various parts of 
Airflow (and various ecosystem components). There is no "one is better than the 
other" - they simply have different characteristics. And sometimes even those 
who provide the "option" or "alternatives" keep it as "ecosystem" part becaus 
then they want to keep their own pace of changes and develop it futther 
(example - user-community chart).
   
   For the Apache Airflow community - following the Apache Way 
http://www.apache.org/theapacheway/ - any new option like that to support makes 
more combinations to test, develop and make sure it works fine for the users. 
That's why anything that we accept as contribution MUST at least have automated 
tests, and we have to be sure it brings value that offsets the cost connected 
with maintaining it.
   
   Also when software brings dependencies we must be sure we are ok with the 
dependencies - licence wise, stabiity wise, future-compatibility and 
maintenance wise. In this particular case Quarkus (which I first hear of) is a 
Java-based dependency which we have almost no experience with as a comunity. 
Any fixes/problems there **might** require skills that are a bit outside of our 
domain. 
   
   I do not know if those are serious problems or and blockers or not  - I am  
just pointing out the areas that are potentially problematic, and that such 
"Add new option for users" might have some serious  implications that we should 
consider.
   
   Accepting a "feature" or "code" is more often than you think "liability" 
rather than "asset".
   
   I am not telling it's the case here, Certainly if we go in this direction, 
comprehensive set of unit tests is absolute MUST - without it, it is "no-go" no 
matter what.
   
   So I think we should wait for others to chime-in (but it might take days or 
weeks depending on how busy they are) - so crertainly do not expect immediate 
actions here.


-- 
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: [email protected]

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


Reply via email to