Access to the Source Code of the Programs You're Using Matters (Even If You're Not a Coder and Cannot Fix Bugs)
The "father of GNU/Linux" always speaks of collective control as a route towards freedom or users' groupwise emancipation:
Laurenz Albe has a new article ("Open source and its power demonstrated in PostgreSQL") which speaks of how access to the full source code and commit history access (in the PostgreSQL project, along with old mailing lists' archives) enabled him to identify - by actual name - a culprit and then fix an old bug. To quote this lengthy article from Albe:
Since I work with the PostgreSQL source code a lot, I already had it on my machine. But it is not hard to get the source code. Appendix I of the PostgreSQL documentation points you to the PostgreSQL Wiki. There we learn how to clone the PostgreSQL source code repository. All you need is to install Git.[...]
In order to understand the motivation behind the change, it is a good idea to read the mailing list discussion. The PostgreSQL project archives the mailing lists, so we should be able to find the discussion. Note that finding the discussion has become easier today, because it has become a habit to reference the relevant mailing list discussion in the commit message.
I used the advanced archive search for the best results. I narrowed the search down to the “pgsql-hackers” mailing list, searched for “autovacuum aggressive”, set “Date: anytime” and “Sort by: Reverse date”. This produced plenty of results, but on the second page I already found this thread from August 2006, which must be the correct one.
To sum up the arguments: Rod Taylor argued that the scale factor should become more aggressive and suggested a value of 0.1. Matthew T. O'Connor and Jim C. Nasby supported that, and Jim suggested a value of 0.2. ITAGAKI Takahiro suggested a value below 0.1 to reduce index page splits. Peter Eisentraut proposed a value of 0.08, which seemed too big a jump to Matthew T. O'Connor. Peter suggested 0.2 as a compromise. Josh Berkus considered that too aggressive and wanted to stay on 0.4. Finally Bruce Momjian settled the dispute by simply committing Peter's proposal.
Companies like Microsoft tell us that full access to all the code isn't important (while they work with the NSA to booby-trap programs with back doors), but this is clearly untrue. If a program is defective, then someone out there might fix it for everyone and if you're not a coder and wish to change some behaviour, then you can hire a programmer (for as little as a few hours) to do this for you. Albe has been working on this program for nearly 2 decades, so he's familiar with the code, the people, etc. █