I wasn't sure exactly where to put this issue. If someone has a better idea of where to put it then let me know but since it's related to a system that only exists on d.o. it made sense to me for me to raise it here. I want to encourage some discussion outside of blogs and meetups, hopefully even inspire some improvements :)

The current system we have is that when a user wants to upload a module as "official" contrib to d.o. they have to go through an application process.

The application process is really well documented and all new modules are forced to go through a few key and basic checks, both peer reviewed and automated including:

- Is this module a duplicate or share a lot of scope with an existing module? could this module be better implemented as a patch? could the applicant achieve their goals better by applying for co-maintainer or maintainer status of an existing project?
- Does this module adhere to basic coding standards and have no "obvious" security issues (this one is largely automated)?
- Does the module work as advertised, or is it missing key functionality or downright bug ridden?
- Does the module leverage the Drupal API wherever possible or is it "re-inventing the wheel" too much?

This all makes sense, but the application process itself changes over time - Automated testing tools are constantly being improved, the coding standards and Drupal API changes and new things are becoming "best practise" all the time. Not just big, obvious things but little things like naming conventions change too that can be hard to stay across until somebody else alerts you to it.

One problem with the current system, when we allow our standards to change (improve) over time, is that we only ask users uploading contrib modules with their own namespace (from the perspective of a new user read "officially sanctioned by the community") to adhere to it the first time they upload *any* module. Once you have git vetted access you can push whatever you want up. The whole thing goes from being a regulated, formalized process to an honesty system with little visibility or QA around new developments.

I see modules currently on d.o., some very popular with hundreds or even thousands of users, that if they were written by a new user today probably would not pass the application process without some serious cleanup. Sometimes the issues are "just" cosmetic (500+ issues flagged by Ventral) but often the modules are duplicating functionality or abusing the Drupal API. I've even seen "modules" that have no downloadable, functional release and seem to exist just so the writer can "reserve" the namespace for what is essentially still a sandbox level project.

I have a few ideas as to how this happens, I'm sure there are many others:

- Developers don't always ensure that every module they promote to contrib status is of the same quality as their first module. Possibly not even getting it peer reviewed before promoting it.
- Contributors are doing their best to make sure their work is always "best practise" but the state of the art and community standards change before they realize.
- A contributor's first application could be much less ambitious than future projects
- Application reviews at some point in the past could have been less stringent than applications in the present

The main point I'm making is not to be dismissive or disrespectful towards the hard work of so many people, but simply that it's not OK to assume that because a person creates work of a certain standard once (maybe only after a lot of help from their peers) that everything they create henceforth will be of equal or greater quality. There's just too much variation over time in any one individual's motivation, abilities and life situation.

A secondary point to requiring experienced developers to rejoin the application and review process through the bonus program would be an immediately and greatly improved review and review bonus program. I know that a common criticism of the current application process is that it feels a lot like the blind leading the blind, at least during the initial stages and there's not nearly enough participation from people who have been successful in the past.

Comments

tvn’s picture

Project: Drupal.org customizations » Drupal.org site moderators
Version: 7.x-3.x-dev »
Component: Miscellaneous » Project/Git problem

Drupal.org customizations queue is for specific code implementations, Webmasters queue looks more suitable. But generally best place for discussions is g.d.o. Particularly groups:
http://groups.drupal.org/peer-review
http://groups.drupal.org/drupalorg

You can find some relevant discussions there, e.g. http://groups.drupal.org/node/139754

Personally I think that without fundamental changes to the project application process (and firstly it's speed), it's unrealistic to require all modules to go through it.

Drave Robber’s picture

Background: this was probably triggered by the discussion on this blog post; even if it was not, there are some interesting and relevant opinions there.

thedavidmeister’s picture

@Drave Robber - Yeah triggered by the blog post, but also something that's been bugging me and I've been discussing "offline" with other Drupalers I know for over 6 months now... since about a month before my own application went through.

Basically, I read that blog post and was like "Wow, I've seen so many people complaining about this process and maintainers asking for feedback a lot" and figured that I should try to get it "in the proper channels" if I could.

@tvn - thanks for moving this issue to where it should be. What I'm suggesting is that if the "Review bonus" program is no longer optional (not really this issue though, that could be discussed separately) and everyone who wants to submit a new application has to go through it then the speed should increase overall as there will be more participation. I see the argument that "we can't get more people to use it until more people use it" as a bit chicken or the egg.

I personally wouldn't support any move towards reducing the minimum quality of the code that is approved for the sake of improving the speed of lower quality applications. I really want to improve the speed and quality of applications together and I think that people who have successfully been through the process are much better equipped to achieve this than an ever changing guard of "first timers".

thedavidmeister’s picture

I had a look over http://groups.drupal.org/node/139754 again - I'm pretty sure I've actually read through it a few times before for different reasons.

While it is related and anybody who comments here should have a go at reading through it all (good luck), it is open ended discussion threads like that which are exactly what prompted me to open an issue. Even the OP of that thread stopped participating in a productive way mid-August 2011.

This is intended to be treated as a specific "issue" with the goal to reach a consensus on this particular idea as "wont fix", "postponed" (hopefully not postponed) or "fixed", not an open ended discussion about the problems, perceived and real, with the application queue in general.

dddave’s picture

Status: Active » Closed (won't fix)

This proposal hasn't gained any interest in over one year and therefore I am closing it as part of the 2014 spring queue cleanup.

My personal 2 cents: I neither think we have the resources to make such a change nor do I think it would be a wise move even if we had them. Instead we would move more and more user away from drupal.org as a module publishing platform.