GNU Emacs Development Inefficiency (2010)

By Xah Lee. Date:

Posted a bug about a problem in minor modes. bug#6611 However, it got closed within a min it was published, and wrongly!

emacs bug 6611 ftbxp
emacs bug 6611 [bug#6611]

It got closed right away i suppose partly has to do with my unforgiving nature of criticizing and run-in with some GNU emacs developers in gnu.emacs.help and comp.emacs in the past 5 or so years.

It's quite frustrating trying to contribute to GNU Emacs. In the past 3 years, i've submitted some 50 bug reports i think by now … without looking at the records, i think at least 10 or so are hard bugs that got fixed. Some of my outspoken criticisms, had their effects, and i suppose some emacs 23's UI changes are influenced by my criticisms. (e.g. line-move-visual, text selection highlight by default, and i noticed yesterday that emacs 23.2.1's doc now removed the phrase “real-time display editor”, which was a item i criticized in Problems of Emacs Manual.)

I've also written to Richard Stallman a few times in private in about 2008 or 2009, about documentation improvements. With extreme politeness and respect on my part. Without going into detail, i'm just disenchanted by his reaction. In short, it appears to me he did not pay much attention, and basically in the end asked me to submit changes to him. Yeah right. The whole shebang seems to be very well described by Ben Wing. [see GNU Emacs and XEmacs Schism, by Ben Wing] (Richard Stallman's emails are pretty short, just a couple terse sentences; but he does, however, whenever he got a chance, tell his correspondents to use the term GNU/Linux, and ask them to contribute.)

Re-writing the whole doc in a modern perspective might take me one month full time. (For example, 160 hours) But if it were to be done in a public way, or submit to him, the time it takes to communicate, email, write justifications, create diffs, etc, can easily take half a year full time (960 hours). In the end, i'm not even sure half of the text in the new doc would be accepted.

The GNU Emacs's bug database sucks. (it's using Debbugs.) The web interface/presentation is incomprehensible. I have problem finding all bugs posted by me or by a bug number. The several times i used it, i still haven't figured out how to find a bug given its bug number. (haven't really tried hard, but to search by bug number or by poster should be something one would intuitively know how to do within 30 secs of browsing the site.) They did not have a bug database, only in around 2008. Most commercial software have a bug database system in 1990s, and most large open source projects have one by early 2000s. (I wrote a bug tracker in 1998, 4k lines of Perl (with CGI, MySQL), in about 2 weeks, for a startup brainpower.com.)

Am pretty sure there are several good “FSF Free” bug databases. [see Comparison of issue-tracking systems] Few years ago, some may have problem to be politically qualified to be “Free” for FSF to adopt. However, these days there are many that FSF officially sanctions as “Free”. However, when you look at FSF, you see that even when a software became FSF Free, they usually are still picky with lots qualms, and typically always ends up using their OWN ones (i.e. from GNU project), even though it is clear that it is inferior. (the GNU emacs dev's revision control system was CVS up to ~2009. CVS has been phased out in early 2000s in majority of software orgs or projects. (GNU emacs now uses Bazaar (BZR). since 2010.)

These are consequence of old and large orgs, with its old policies and bureaucracies. See: “Free” Software Morality, Richard Stallman, Paperwork Bureaucracy.

Who are the main developers of FSF software these days? Mostly, they are either paid as FSF employee, or students still trying to break out their craft in programing, or 40 or 50 years old semi-retired programers who otherwise isn't doing anything. Those willing and able, spend time and get decent salary in commercial corps, or went to start their own projects or business that'd be far more rewarding financially or recognization than being another name in FSF's list of contributors.

These days, FSF and Richard Stallman more serves as a figure-head and political leader in open source movement. FSF's software, largely are old and outdated (For example, unix command line utils), with the exception of perhaps GCC and GPG. If we go by actual impact of open source software in society, i think Google's role, and non-profit orgs (such as Apache, Perl, Python, PHP, Ruby, various languages on JVM), and other commercial orgs (For example, Apple, and other commercially founded project hosters for hosting open source projects (Google Code, SourceForge, github, etc.)), exceeded FSF by early 2000s.

Thanks to Alan Mackenzie [http://www.emacswiki.org/emacs/AlanMackenzie] for correction on revision control system used by GNU Emacs.

See also: Emacs Dev Inefficiency and Emacs Web 2.0? (2010).