Yes, it's hard to know what can conclusively be drawn from the raw totals.
I do think trends and ratios (e.g. Py2 vs. Py3) will, however,
roughly reflect underlying usage (which itself is ambiguously defined).

On Thu, Apr 9, 2020 at 7:30 PM Kenneth Knowles <[email protected]> wrote:

> Yea, interpreting the raw absolute number is tricky. You can probably
> manage to see certain kinds of trends if you just look at relative numbers.
>
> Kenn
>
> On Thu, Apr 9, 2020 at 6:42 PM Austin Bennett <[email protected]>
> wrote:
>
>> @Robert Bradshaw <[email protected]>  , you sent that pypi link [1]
>> the other day in response to something else, which is what prompted me to
>> ask Gris about Maven (based on that link [2], @* Kenneth Knowles
>> <[email protected]>  ).  I recall talking to someone about Maven download
>> statistics at ApacheCon.
>>
>> Perhaps these are not the only sources; happy to explore any/all that
>> others may have in mind.  Generally (personally) interested in
>> understanding our ecosystem, trends, etc.
>>
>> Naturally, there is a ton open to interpretation, even if we had absolute
>> raw data (down to IP/account/etc, which would also have ManyIps to one
>> user, as well as many users to one IP, in the case of uses on corporate
>> networks).  Getting distinct users, for instance, would be incredibly
>> challenging; not sure how/whether other projects even do such things.
>>
>> [1] https://pypistats.org/packages/apache-beam
>> [2]
>> https://blog.sonatype.com/2010/12/now-available-central-download-statistics-for-oss-projects/
>>
>> On Thu, Apr 9, 2020 at 4:02 PM Kenneth Knowles <[email protected]> wrote:
>>
>>> I found some info from 2010 [1] that it was available to anyone with
>>> deploy permission. The instructions still work.
>>>
>>> Kenn
>>>
>>> [1]
>>> https://blog.sonatype.com/2010/12/now-available-central-download-statistics-for-oss-projects/
>>>
>>> On Thu, Apr 9, 2020 at 3:41 PM Robert Bradshaw <[email protected]>
>>> wrote:
>>>
>>>> For Python, there's https://pypistats.org/packages/apache-beam . It's
>>>> unclear how accurate these are, and how many of these downloads represent
>>>> users vs. tools (e.g. setting up environments for continuous testing).
>>>>
>>>> On Thu, Apr 9, 2020 at 3:29 PM Griselda Cuevas <[email protected]> wrote:
>>>>
>>>>> Hi folks - I'm interested in knowing more about Beam's adoption
>>>>> through user downloads.
>>>>>
>>>>> Do you know what's the protocol to access Maven and check on Java
>>>>> downloads?
>>>>>
>>>>> Also - do you have any other recos on how to measure the project's
>>>>> adoption evolution?
>>>>>
>>>>> Thanks!
>>>>> G
>>>>>
>>>>

Reply via email to