[[[ To any NSA and FBI agents reading my email: please consider ]]] [[[ whether defending the US Constitution against all enemies, ]]] [[[ foreign or domestic, requires you to follow Snowden's example. ]]]
> For many years GNU projects like Coreutils have > stopped maintaining ChangeLog files by hand: the files are still placed into > distribution tarballs to satisfy the GNU coding standards, but are autogenerated > from Git commit messages, and developers don't read them. This has worked well. Would you please show me what those automatically generated change log files for Coreutils look like for the past 3 years? Then I could check directly what information they contain, and whether anything I consider important has been left out. > Of course, git commit messages can be done poorly, just as ChangeLog files can > be done poorly. Quality of writing of the log messages is important, but that's not the issue this discussion is about. The issue what information they contain. How the information is expressed, and where it is stored, are not crucial issues. I'm not wedded to the current format, or having this all in a file called ChangeLog. I am concerned about making sure that future developers can get at all the information that properly written ChangeLog files have made available thus far. Many people seem very eager to get rid of ChangeLog files. This eagerness leads me to be concerned that they are too quick to conclude that nothing important would be lost. Perhaps that is true, but I want to make sure of this. I am not looking for an excuse to say no. I would be glad if we solve these problems and then drop hand-written ChangeLog files. But pressure and repetition won't convince me. What will convince me is a demonstration. -- Dr Richard Stallman President, Free Software Foundation (https://gnu.org, https://fsf.org) Internet Hall-of-Famer (https://internethalloffame.org) Skype: No way! See https://stallman.org/skype.html.
