November 17, 2015
On Monday, 16 November 2015 at 15:20:51 UTC, Andrei Alexandrescu wrote:
> http://erdani.com/d/downloads.daily.png
>
> There have been 1677 dmd downloads per day (net after discounting Travis CI) on average over the past 28 days (i.e. four weeks ending Sunday, November 15).

Hello Andrei,

what do you think how good the download numbers are representing the popularity of D? Because I myself have downloaded the new compiler several times. One for work, one for home and one for the virtual machine I guess.
November 17, 2015
On 11/17/15 8:08 AM, Namal wrote:
> On Monday, 16 November 2015 at 15:20:51 UTC, Andrei Alexandrescu wrote:
>> http://erdani.com/d/downloads.daily.png
>>
>> There have been 1677 dmd downloads per day (net after discounting
>> Travis CI) on average over the past 28 days (i.e. four weeks ending
>> Sunday, November 15).
>
> Hello Andrei,
>
> what do you think how good the download numbers are representing the
> popularity of D?

Your guess is as good as mine. It's just a proxy. Generally more daily downloads indicate an increasing interest. -- Andrei
November 17, 2015
On Tuesday, 17 November 2015 at 13:08:37 UTC, Namal wrote:
> what do you think how good the download numbers are representing the popularity of D? Because I myself have downloaded the new compiler several times. One for work, one for home and one for the virtual machine I guess.

Oh the other hand, you have people like me who often skip new downloads but use D all the time anyway, and people who get them through third party package managers, etc.

My gut feeling is that it probably basically balances out, so more downloads probably means more users, though we couldn't actually tell how many users by just looking at this.
November 17, 2015
On Tue, Nov 17, 2015 at 06:42:37PM +0000, Adam D. Ruppe via Digitalmars-d-announce wrote:
> On Tuesday, 17 November 2015 at 13:08:37 UTC, Namal wrote:
> >what do you think how good the download numbers are representing the popularity of D? Because I myself have downloaded the new compiler several times. One for work, one for home and one for the virtual machine I guess.
> 
> Oh the other hand, you have people like me who often skip new downloads but use D all the time anyway, and people who get them through third party package managers, etc.
[...]

And I never download D from dlang.org; I pull from github. Of course, only a very small subset of D users would do this. :-P


T

-- 
Ph.D. = Permanent head Damage
November 17, 2015
On Tuesday, 17 November 2015 at 13:08:37 UTC, Namal wrote:
> On Monday, 16 November 2015 at 15:20:51 UTC, Andrei Alexandrescu wrote:
>> http://erdani.com/d/downloads.daily.png
>>
>> There have been 1677 dmd downloads per day (net after discounting Travis CI) on average over the past 28 days (i.e. four weeks ending Sunday, November 15).
>
> Hello Andrei,
>
> what do you think how good the download numbers are representing the popularity of D? Because I myself have downloaded the new compiler several times. One for work, one for home and one for the virtual machine I guess.

As long as we didn't change something in D that affects how often one person downloads the compiler, these are independent variables and do not affect the trend. One or three years ago (or if D were as it was one or three years ago), would you not have downloaded the compiler the same number of times?
November 17, 2015
On Tuesday, 17 November 2015 at 23:26:15 UTC, Vladimir Panteleev wrote:
> On Tuesday, 17 November 2015 at 13:08:37 UTC, Namal wrote:
>> On Monday, 16 November 2015 at 15:20:51 UTC, Andrei Alexandrescu wrote:
>>> [...]
>>
>> Hello Andrei,
>>
>> what do you think how good the download numbers are representing the popularity of D? Because I myself have downloaded the new compiler several times. One for work, one for home and one for the virtual machine I guess.
>
> As long as we didn't change something in D that affects how often one person downloads the compiler, these are independent variables and do not affect the trend. One or three years ago (or if D were as it was one or three years ago), would you not have downloaded the compiler the same number of times?

package manager presence has improved, so I would expect dlang.org downloads to represent a smaller fraction of total downloads than it used to.
November 18, 2015
On 2015-11-18 00:26, Vladimir Panteleev wrote:

> As long as we didn't change something in D that affects how often one
> person downloads the compiler, these are independent variables and do
> not affect the trend. One or three years ago (or if D were as it was one
> or three years ago), would you not have downloaded the compiler the same
> number of times?

Personally I have more machines now to download the compiler to, supporting more platforms.

-- 
/Jacob Carlborg
November 18, 2015
On Monday, 16 November 2015 at 15:20:51 UTC, Andrei Alexandrescu wrote:
> http://erdani.com/d/downloads.daily.png
>
> There have been 1677 dmd downloads per day (net after discounting Travis CI) on average over the past 28 days (i.e. four weeks ending Sunday, November 15).

A moving average is probably the most overall useful graph, but would it be possible to also have a graph without a moving average (i.e. simple daily tallies)? That would make it easier to pinpoint individual days of high activity, e.g. due to media coverage.
November 18, 2015
On Wednesday, 18 November 2015 at 08:22:19 UTC, Jacob Carlborg wrote:
> Personally I have more machines now to download the compiler to, supporting more platforms.

Isn't this a proof that it is expanding?
November 18, 2015
On 11/18/2015 04:00 AM, Vladimir Panteleev wrote:
> On Monday, 16 November 2015 at 15:20:51 UTC, Andrei Alexandrescu wrote:
>> http://erdani.com/d/downloads.daily.png
>>
>> There have been 1677 dmd downloads per day (net after discounting
>> Travis CI) on average over the past 28 days (i.e. four weeks ending
>> Sunday, November 15).
>
> A moving average is probably the most overall useful graph, but would it
> be possible to also have a graph without a moving average (i.e. simple
> daily tallies)? That would make it easier to pinpoint individual days of
> high activity, e.g. due to media coverage.

Plotting daily downloads looks uninformative because of the high variance. But we could publish data as tabular information. Making the stats script available is on my list. -- Andrei
1 2 3