I sent Chris the following e-mail today and he asked that I post it to the full community. John Brice Meadville Public Library Dear Chris:
Hi! John Brice here from Meadville.
I usually just lurk about these mail lists but I thought I would give you my two cents worth concerning the sign off process.
First, we seem to have three issues that I can identify.
Issue 1. There is a large number of submissions (140 at last count) that need to be signed off on.
Issue 2. A small number of the patches our very specific and are difficult to test unless one is running the exact same configuration.
Issue 3. Quality assurance is becoming more and more critical especially considering that the installed base is getting bigger. One small problem in the code can now effect 1,000's of libraries across the world.
Really, to solve the above three issues you have to identify the most important issue and then work backwards from their to come up with the proper solution from there.
I personally believe that the Issue 3 Quality Assurance is the most important. There are just too many libraries out there now relying on Koha, in a production setting, to risk putting in code that has not been properly vetted. The bottom line is that only a low percentage of libraries will benefit from a new feature while everyone benefits from clean code.
Having said that how do we fix issue One and Two? Well Issue one seems to be a manpower issue. While, Issue two is a very difficult technical issue concerning system configuration.
The best way to solve Issue Two is to have a waiver process for certain developers. Certain developers associated with a large percentage of code development could, in certain specific circumstances, request a waiver from the traditional sign off process. The reasons for the waiver would have to be very specific, such as it would be too difficult too test outside of a production server. The waivers would have to be tracked (yeah another thing for the database to keep track of) and if there is any problem at all the entity granted the waiver would have to be responsible for all subsequent revisions and fixes. The waiver is not the solution to reduce the large number of outstanding issues it is a simple means to add code to Koha that is too difficult to test outside of a very specific configuration.
So then we come to the big Issue Number 1 the large amount of checks that have to be made. This is the problem with open source we have to rely on the good graces of everyone involved in order to move the project forward. This means that everyone has to set aside their own personal projects and do stuff for the good of the overall Koha project. That is a tough sell, but frankly, it is one that is needed. The best way to get this logjam cleared up is to have someone be put in charge of the issue and then that person start writing e-mails and so forth to encourage cajole or outright bang heads to get the sign-offs that are required. If all the developers in Koha would sign off on one outstanding issue a week we could have the problem resolved in three to four months. Someone needs to keep hammering away at this issue and I don't think it should be you Chris.
Just some thoughts from a librarian stuck in the Allegheny National Forest.
John Brice