[Python-Dev] PyPI comments and ratings, *really*?

[Michael Sparks]

On Thu, Nov 12, 2009 at 2:06 PM, Jesse Noller <jnoller at gmail.com> wrote:
...
> Frankly, I agree with him. As implemented, I *and others* think this
> is broken. I've taken the stance of not publishing things to PyPi
> until A> I find the time to contribute to make it better or B> It
> changes.

Ditto, but maybe for different reasons.

Personally, I'm interested in feedback - good and bad. That's the
reason I choose odd names for projects, since it means I can create a
google alert to find out random comments in bizarre locations (hence
why when you wrote a blog entry, I responded). However, the reason I
released anything onto PyPI (and relunctantly at that) was due to a
random complaint that a user couldn't go "easy_install <foo>" and have
it pick up code from PyPI.

Going along with comments made elsewhere (by Guido I think) saying
"but user's like reviews and rating when someone publishes a book",
probably using Amazon, B&N & similar as examples, I agree they do.

The closest equivalent here though IMO is somewhere like lulu.com -
where people self-publish. Like PyPI that has a ratings system and
comments, so you could say if it "works" there it should work for
PyPI.

The problem though is that software is much more mutable that a book.

Taking the example listed - a comment added here:
  http://pypi.python.org/pypi/spypam/1.0

There's a note:
   "Inadequate docstrings give no clue about function arguments. Dumps
core when I call it after reading the source to figure out the API.
Cannot recommend."

That's useful from a user perspective. Or is it? It's useful from a
user perspective, until that issue is fixed. Then what? Is it still
useful? Can the commenter remove it? Can they get notified it's
changed? Can the maintainer say "this is fixed/changed?"

I never look at the PyPI pages for stuff I create. Which means if
someone is using it for support, they're wasting their time. (Why
would I look at it? I know what the project is for and where to get
it! :) (and also PyPI isn't the prime download for it either - so the
download stats are irrelevant to me) I doubt I'm alone, so how many
people's time are wasted asking questions there ?

*shrug*

I suppose, personally, I'm dubious about the idea of having unchanging
comments and ratings associated with projects which are changing and
improving - that feels like a mismatch. (Unlike a book, which
generally is unchanging or has a separate edition and separate set of
ratings and reviews)

Incidentally, and perhaps probably more relevant to the discussion
than my random opinion - some time back I created the twitter id
http://twitter.com/pypi - using twitterfeed - since I wanted an easier
way of following additions to pypi. There's currently 774 people
following that.

If there's interest, and if there's a survey to be done, I could
forward a link to a survey through that twitterfeed - which I suspect
is a mix of users of PyPI and uploaders to PyPI.

(On a secondary note, if there's someone else who thinks they should
own it, please let me know - it was a random convenience that people
seem to find useful :-)


Michael.