On Mon, 2 Jul 2018, Dennis Clarke wrote:

> It may be of value to create a curl/libcurl "policy" wherein the annual
> survey data may indicate the degree to which a feature is an opt-in or a
> default feature. In this way one would be able to say that there is less
> than a 5% usage rate for a given feature and thus it is now an "opt-in"
> component. That avoids the entire discussion and also renders the project
> policy as something well understood. There may be other such "features"
> that are barely used at all but adding to the size and complexity of a day
> to day essential tool and lib. Just my thoughts.

I like that idea! But I don't think we can use that to avoid me having to send
emails like this because...

- the user survey is unreliable as the ultimate "truth" since it is only a
   small subset of users filling it out

- I still want to explain my thinking and I hope someone will tell me if I
   forgot something important that makes my plan break or just be bad. If
   nobody says anything, that means nobody objects or found a serious problem
   with the plan (that they can't live with).

- sometimes a "single user" of a feature uses libcurl in an application
   used by a billion humans. Such use cases might get some extra attention and
   sympathy.

> As for http/0.9, feels like it is below the low watermark but this is just a
> feeling.

I also use these emails (when archived) as something to point to in the future
when someone crawls out of the shadows and is mighty upset because we removed
a feature they NEED...

-- 
  / daniel.haxx.se
-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette:   https://curl.haxx.se/mail/etiquette.html