Opinion poll: feedback URL upon import

[Stéfan van der Walt]

Hi all,

It has been difficult to discover use cases of scikit-image out in the wild (even though I know they are plentiful, and users would excitedly tell me about them at conferences once the topic is broached).

So, I was wondering if it would be considered bad form to print a message upon importing scikit-image, something along the lines of:

"Thank you for using scikit-image v0.12! Please help us by telling us about your use-case at http://scikit-image.org/feedback?v=0.12.3"

Thanks for your feedback,

Stéfan

[Emmanuelle Gouillart]

Hi Stéfan

I'm quite against it, since it's a feature that would push scikit-image
to be more an application and less a self-standing library: we would like
any other package to be able to import skimage in their code without
bothering about such messages. Indeed, such a message at import is only
useful/relevant for people that use scikit-image deliberately and as an
important part of their workflow, not so much for people that import
skimage in another projects, or that just use a couple of functions of
scikit-image (but it's not the core part of their workflow). I consider
it important for scikit-image to be a versatile block of the ecosystem,
and not only an application at the end of the "food chain".

We could advertise the feedback form in a different way, for example by
making it very visible on the website. With 20000 unique visitors coming
to the website every month, I hope that we could get some feedback...

Best,
Emma

[Christoph Deil]

Hi,

I agree with Emma, as a scikit-image user, I would find it annoying.

Especially if it catches on and many other Python packages start doing the same.

It’s probably very rare, but if people check the text output of their library or application and use scikit-image somewhere, this could even break their test build.

In Astropy there has also been some discussion about how to get feedback from users. There the discussion was more along the lines of wanting to collect info which platform, versions are in use, and which sub-packages and functions in the Astropy package are used. I.e. there was discussion about adding some code to send such usage stats info to a server, like many apps do. This didn’t happen, I think the idea has been dropped.

It would be very nice and important to get more feedback from users!

But it seems to me that asking for it on the website like Emma proposes or in conversations and presentations might be the safer way.

Christoph

-- 
You received this message because you are subscribed to the Google Groups "scikit-image" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image...@googlegroups.com.
To post to this group, send an email to scikit...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/scikit-image/20160426064948.GB940933%40phare.normalesup.org.
For more options, visit https://groups.google.com/d/optout.

[Robin Wilson]

I agree with the others: please don't do this!

However, being more obvious about wanting feedback, via the website, mailing list, in conference presentations etc, is definitely a good idea.

Robin

[Jaidev Deshpande]

To post to this group, send email to scikit...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/scikit-image/55d7ee35-aae3-458f-8245-1d35756dd0ce%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

How about posting this message when skimage is installed? It could be added to the setup.py script. But that would be visible only for users installing skimage from source, since pip / conda / enpkg would most probably suppress the message.

Would it make sense to add a message in the LICENSE file?

--

JD

[Kevin Keraudren]

Noisy output at import is bad. In addition to the suggestions previously made, what about also mentioning the feedback url in the main docstring of skimage, something like here: https://github.com/scikit-image/scikit-image/blob/master/skimage/__init__.py#L9

To view this discussion on the web, visit https://groups.google.com/d/msgid/scikit-image/CAB%3DsuEkN-VdjQ%3DZ6%3DRXKnMX2mwotjqVJ2X3HO1ihRH8k-iDQ8w%40mail.gmail.com.

[Juan Nunez-Iglesias]

-1, as others. The suggestions given so far are good: prominent message on the website, and in the main docstring.

Juan. 

To view this discussion on the web, visit https://groups.google.com/d/msgid/scikit-image/CADe6jSp756gGkgXYhdjvGzzD2q3g2-y9%2BiK%2BSEWQpOetq1pWaQ%40mail.gmail.com.

[Stéfan van der Walt]

I guess this thread makes it clear what I should do:  threaten to include such a message unless people send me usage reports :)

Stéfan

To view this discussion on the web, visit https://groups.google.com/d/msgid/scikit-image/CA%2BJHcKTn6zehuanT6ZqBhkd1W5LujxV%3D7CNbhUfTmPHjm%2Buf3w%40mail.gmail.com.

[Andreas Mueller]

There was some discussion about whether we could use jupyter to get some feedback from users, i.e. a single opt-in for the jupyter
platform, instead one for each library (which would be pretty bad).

One method that I recently used was greping notebooks on github. Unfortunately their API doesn't allow
global code searches. Their website does, though, if someone wants to write a scraper ;)

I wonder if / how other open source libraries handle this. I guess boost doesn't care ;)
But maybe web frameworks do? We could try tapping other scientific computing communities like
julia or octave, or go out to web folks?
The scipy stack is probably not the only one that has this issue.

To view this discussion on the web, visit https://groups.google.com/d/msgid/scikit-image/CABDkGQnwf%2BiuidPM%2BCf9fjQRQOw03UGR8ih1hG%3DKyoObP-4koQ%40mail.gmail.com.

[Stéfan van der Walt]

On Fri, 29 Apr 2016 at 18:54 Andreas Mueller <amue...@ais.uni-bonn.de> wrote:

There was some discussion about whether we could use jupyter to get some feedback from users, i.e. a single opt-in for the jupyter
platform, instead one for each library (which would be pretty bad).

I think this is still on Nathaniel's radar; he's been quite busy with the manylinux work, but now that that's almost done I guess he'll round back on this project.

 

I wonder if / how other open source libraries handle this. I guess boost doesn't care ;)
But maybe web frameworks do? We could try tapping other scientific computing communities like
julia or octave, or go out to web folks?

I had another good suggestion from a colleague yesterday that we can have a showcase section on the website, where authors & companies can get free publication for work using scikit-image.

I also want to add a survey bubble to the website, to gauge where we can best improve.

Stéfan