TIOBE December 2015 - D rose 5 positions

[Joakim via Digitalmars-d]

On Sunday, 3 January 2016 at 17:25:13 UTC, Ola Fosheim Grøstad 
wrote:
> On Sunday, 3 January 2016 at 16:56:46 UTC, Joakim wrote:
>> It's more than not being neutral: I pointed out that github 
>> suffers from similar categorization errors to the ones you 
>> list below.  But yes, github stats are really only good for 
>> languages used in open source, and OSS is still a small 
>> fraction of all software written.
>
> I didn't notice any miscategorization in the trending lists I 
> used, there might have been some, but even then the surrounding 
> numbers were similar at #20 of the monthly trend so I don't 
> think that the numbers would be significantly effected for most 
> languages.

I gave you a specific example, the 32nd most starred "D repo" 
according to github, which has nothing to do with D (there are 
several more miscategorized like that, look at #22 in the above 
list).  The amazing part is that if you click on the D link on 
the language breakdown for that cocos2d repo, it'll say it 
couldn't find any D files!  So it somehow miscategorizes that 
project as primarily built on D, despite github's filename 
analyzer not finding any D files. :)

> Open source is a good indicator of trending. It is also a 
> indicator for the productivity of new languages when a new 
> language produce many popular open source applications. It is 
> also a good indicator of what programming areas a new language 
> is going to be popular in. If a language has a long tail of 
> popular applications, that's a pretty strong indicator that it 
> will take off, IMO. Go is there. Rust isn't quite there (yet). 
> D and Nim doesn't show any signs of going there (yet).

Those are good hypotheses, not sure you can say OSS usage is a 
good _indicator_ yet, especially since I wouldn't say Go has 
taken off.

>> Are you stating that Docker is built on Go or suggesting it 
>> would make sense if they were?  Sounds like the latter.
>
> Docker is based on Go.
>
> «Under the hood, Docker is built on the following components:
>
> The cgroups and namespaces capabilities of the Linux kernel
> The Go programming language
> The Docker Image Specification
> The Libcontainer Specification»

You seem to have gotten that quote from the README for the OSS 
project, but we were talking about the stack at Docker, the 
company.  I highly doubt that OSS github repo is all they have to 
their stack, as they would have tons of competition and wouldn't 
be able to garner that ridiculous private valuation if they did, 
so my question was if they've talked about whether Go also powers 
the rest of their stack.

>> for the long-term future of D.  Better to focus on making the 
>> best language you can, and people will find uses for its 
>> unique strengths.
>
> Yes, but the trends show that D has been going down over the 
> past 8 years and has been stagnant over the past 2 years if you 
> look at statistics for github and google trends.

I'm not sure what to make of the google trends data, but it's 
possible that much D code is simply not OSS.  Certainly you'll 
find very little of Sociomantic's codebase on github.

> So, D has to make a change. Most likely a language change that 
> makes it easier to deal with. People say that Go and Swift are 
> doing well because they are easy to deal with. I think they are 
> right.

I think D has a real complexity problem, so anything we can do to 
make it simpler _while still maintaining its power_ would be 
welcomed.  But D is a machine shop, it's not a screwdriver like 
Go.  You can't make a machine shop as easy to use as a 
screwdriver, unless you plan on staffing it with robots.  If you 
have a proposal to replace all D programmers with robots, then 
I'd like to hear it. :)

>> Heh, why should I spend any time thinking about it whatsoever?
>
> It doesn't take any thinking to see that Tiobe is bogus! ;^)

It takes time to examine their process, something I don't care to 
do since I don't care what their results are!  If I did, I 
wouldn't be working on D, which has never featured very highly in 
their ranking.

>> all.  Since practically every company has a website that 
>> likely uses a little javascript, that's trivially true, yet 
>> completely irrelevant.
>
> First we need to define what we want to measure. But claiming 
> that the marketshare of Javascript is 2.5% is outragous no 
> matter how we measure it.

It's a measure of buzz, not usage or share, as I said with Obj-C. 
  I could see JS buzz going down in recent years, but 2.5% does 
seem low.

>> If you were able to compile something like billable hours for 
>> javascript, it would do well, but nowhere near the top.
>
> I think it would be on the top now, yes.

You think there were more hours billed by programmers for 
javascript in 2015 than any one of Java, C, C#, or C++?  I doubt 
it.

>> Unfortunately, that key is not available under the lamppost we 
>> have: TIOBE. :)
>
> A broken lamppost without electricity that went out of date 
> over a decade ago.

It would be more succinct and accurate to say that you don't 
think the key is anywhere near that lamppost. :)

>> C++ has been retreating into a niche, along with other 
>> AoT-compiled languages, even TIOBE shows that in its graph of 
>> C++ buzz dropping significantly over the last decade.
>
> C++ is going down slowly, but probably will reach a plateu with 
> enterprises that can afford "large" C++ teams.

Which are also a dying breed.

>> javascript is primarily a frontend language on a single 
>> platform, web apps, it will never get "close to the top" of 
>> the programming language heap.
>
> I wish. Unfortunately this isn't true anymore. Electron, 
> node.js and other frameworks are projecting JavaScript as a VM 
> into more and more application areas.
>
> We are just not following the youngsters. They grew up with 
> JavaScript. They will make it pervasive.

On the contrary, I suspect that we've passed "Peak JS", with 
WebAsm about to cripple it.