Are there any good metrics of which programming languages are most popular for greenfield development, i.e. development that is not significantly constrained by legacy code? IMHO this is the most meaningful definition of popularity, since it tells you what people are choosing when their choices are relatively unconstrained by choices made in the dark and distant past. However, the metrics I know of, such as number of available job openings and the TIOBE index, are deeply affected by the existence of legacy code, and do more to tell about what languages have been popular than what languages people are choosing now, given relatively unconstrained choice.
+2
A:
One possible metric could be Stack Overflow's tags, sorted by popularity. Surely it partly reflects the dark and distant past, but probably people are having more questions on new technologies.
It seems that C# is clearly the most popular greenfield language today. On the other hand, Java, C++ and Python, among others, stay stable and strong.
Joonas Pulakka
2009-02-18 18:22:22
C# and Java here.
Karl
2009-02-18 18:25:38
All this shows is that C# developers have less experience as they have to ask more questions. Ever check out the C# questions?
ng
2009-02-18 18:29:19
Stackowerflow is heavily biased towards .net platform, since its owners are, so it is not a representative sample.
Dev er dev
2009-02-18 18:33:51
It's not _all_ this shows. Of course C# developers have less experience on average, but they wouldn't be asking the questions if they weren't using the language. But surely this is quite a fuzzy problem to give any precise, quantitative answers.
Joonas Pulakka
2009-02-18 18:35:03
@Marko, I don't think the site owners' preferences have much to do with the distribution of content. I didn't even know they're .net fans.
Joonas Pulakka
2009-02-18 18:37:10
People working on legacy systems have questions, too, and new projects are not necessarily "greenfield" in this sense. A new project here would be in C++/MFC or C#/.NET, no matter how tempting it was to write it in LOLCODE or Intercal.
David Thornley
2009-02-18 20:20:24
This is a good example of the fact that a language's success is mainly determined by a) its libraries and b) its ability to co-operate with existing languages. The language itself has much less to do with its success.
Joonas Pulakka
2009-02-19 08:11:08