Not every software product is suited for life at DB. Before proposing a new subproject, it is important to read this document carefully and determine whether your product is a good fit.
Here are some best practices that we will expect to find in a successful proposal. This is not a checklist, but guidelines to help communicate what is expected of a DB subproject.
Meritocracy. Before submitting a proposal, be sure to study the guidelines that govern DB subprojects. These guidelines explain our system of Meritocracy, which is the core of the DB Project. If the product developers do not wish to adopt this system, then they should not donate their code to the Project, and should look elsewhere for hosting.
Community. DB is a Project of the Apache Software Foundation. A prime purpose of the ASF is to ensure that the Apache projects continue to exist beyond the participation of individual volunteers. Accordingly, a prime criteria required of any new subproject is that it already enjoys -- or has a high potential for attracting -- an active community of developers and users.
Proposals for non-established products have been accepted, most often when the proposal was tendered by experienced open source developers, who understand what it means to build a community.
A design document, developer and user documentation, example code, and a working implementation all help to increase the likelihood of acceptance. Well-documented products tend to build stronger communities than products that rely source code and JavaDocs alone.
Core Developers. To considered, a product must have at least 3 active developers who are already involved with the codebase. This is an absolute minimum, and it is helpful for there to more than three developers. It is very helpful for at least one of the developers making the proposal to already be active in a DB subproject or other open source initiative.
At DB, the core developers of a product (the Committers) manage the codebases and make the day-to-day development decisions. We are only interested in products that can guided by their own development communities. DB provides the infrastructure, and some essential guidelines, but the Committers must take responsibility for developing and releasing their own product.
Alignment with existing Apache subprojects. Products that are already used by existing subprojects make attractive candidates, since bringing such a codebase into the Apache fold tends to benefit both communities. Products with dependancies on existing Apache products are also attractive for the same reason.
Scope. DB products are generally server-side software solutions written for the Java platform. Proposals for products outside this venue will have greater difficulty in being accepted.
These are anti-patterns that we have found detrimental to the success of a proposal. Some of these are lesson learned from the school of hard-knocks, others are taken from proposals which have been rejected in the past. All will raise a red flag, making it unlikely that a proposal will be accepted.
Orphaned products. Products which have lost their corporate sponsor (for whatever reason) do not make good candidates. These products will lack a development community and won't have the support needed to succeed under the DB umbrella.
For example, we have seen proposals that contain paragraphs like this:
FooProduct is currently a production quality product, and in fact is being used by a live website. It was developed as a product by FooCompany, with the intention that we would sell it. However, due to various economic factors such as the decline in FooProduct's intended market and the recent difficulties in obtaining venture capital, we have decided that at this time it is not feasible for us to continue in that direction.
The alleged quality of a product is not the prime criteria. To be accepted, we must believe that a product will attract volunteers to extend and maintain it over the long term. A product like this, arriving with no volunteer developers or open source community, does not further DB's mission, and would not be accepted.
We generally recommend that an orphaned product start with an independent host, and consider making a proposal after it has started to build a community.
Inexperience with open source. We often receive proposals that start with "We will open our software if you choose to accept it." This is the wrong way to approach the proposal process. A closed source project does not have an open community, and so we have no way to tell if the developers can work in an open source environment. Products that have lived on their own and have started to develop their own community, have a much better chance of being accepted.
If the product's developers have not worked with open source before, it is recommended that they spend some time contributing to an existing open source project before trying to manage one of their own. Since DB subprojects are managed by their own Committers, it is important that a new product supported by people who understand how open source works.
Homogenous developers. Apache communities attract developers with diverse backgrounds. Products with a tightly-knit team of developers may have difficulty shepherding new developers into the fold. It is vital that we believe the developers will discuss development issues openly with the community, and not make decisions behind closed doors. We are especially wary of products with developers who all share a common employer or geographic location.
Reliance on salaried developers. DB has strong ties to the business community. Many of our developers are encouraged by their employers to work open source projects as part of their regular job. We feel that this is a Good Thing, and corporations should be entitled to contribute to open source, same as anyone else. However, we are wary of products which rely strongly on developers who only work on open source products when they are paid to do so. A product at DB must continue to exist beyond the participation of individual volunteers. We believe the best indicator of success is when developers volunteer their own time to work open source projects.
No ties to other Apache products. Products without a tie to any existing Apache product, and that have strong dependencies on alternatives to Apache products, do not make good candidates. Many Apache products are related through common licenses, technologies, and through their volunteers. Products without licensing, technology, or volunteers in common will have trouble attracting a strong community.
A fascination with the Apache brand. The Apache Software License is quite liberal and allows for the code to used in commercial products. This can induce people to donate a commercial codebase to the ASF, allow it developed as open source for a time, and then convert it back to commercial use. While this would legal under the Apache Software License, we are wary of proposals that seem to be more interested in exposure than community.
Final acceptance is based the rules defined in the Project Management Committee Bylaws which state that "Creation of a new subproject requires approval by 3/4 vote of the PMC." The proposal for a new subproject must submitted to the DB General mailing list, where the PMC conducts business. All discussion regarding the proposal will occur the General list, including the final vote.