[E-Lang] measurement of programming language popularity in the open source community

Zooko zooko@zooko.com
Thu, 18 Oct 2001 07:30:08 -0700

(I, Zooko, wrote the lines prepended with "> > ")

 stevej wrote:

> > My methodology [shouldn't that be "method"? -Ed.] was to look at each
> > programming language and count how many projects had greater than 6%
> > popularity.  I chose 6% because Cygwin has 6.52% popularity and I like Cygwin.
> > Popularity is defined as (record hits * URL hits * subscriptions)^(1/3), as
> > described in the freshmeat FAQ:
> > 
> > http://freshmeat.net/faq/view/30/
> > 
> > C:          67
> > C++:        23
> > PHP:        17
> > Perl:       10
> > Unix Shell:  8
> > Assembly:    5
> > Python:      3
> > Java:        2 (Hm... but they are Mozilla and Netscape)
> > JavaScript:  2 (Mozilla, Netscape)

> I hate to say but this method of yours seems a bit askew; Assembly ranks
> as highly as Python and Java combined! I'm an old hat assembly geek and
> I still can't fathom this.

Yes, it's rather surprising.  I think that this is due to the narrowness of the
sample population (open source).  The popularity contest results, quoted above,
make it even narrower: very popular open source apps.

On the other hand, these surprising numbers are actually telling us something,
at least something about the small and narrow world of open source hackery.  Do
*you* have any Java apps installed on your computer, not counting the JDK
itself and the E interpreter?  I don't.

By the way, if you want a slightly less narrow view, better to look at the
results from "all apps indexed by freshmeat", the top 12 of which are duplicated
here.  Assembly no longer ranks as highly as Java and Python combined.

C:          2673
Perl:       1361
C++:        1020
PHP:         740
Java:        726
Python:      403
Unix Shell:  246
Tcl:         190
SQL:         124
Other:        89
JavaScript:   85
Assembly:     83

I don't think that the world of open source is representative of the larger
world, nor do I necessarily think that the E project should target it in
specific, but it is easier for me to find these kind of statistics about the
open source world than about other arenas of programming.