[Bigbang-dev] Thoughts on plans for future releases

[Sebastian Benthall]

>
> I like the idea of getting more tests in there now in part so that it can
> feel like a more regular practice to add tests whenever we add new features.
>

Looks like we are at 45% test coverage already--not bad!
https://gist.github.com/sbenthall/05f585518cac8e27039a7c1b5da11a62

There's this old ticket calling for 60% test coverage, which seems like a
good target for the 0.2.1 release:
https://github.com/datactive/bigbang/issues/38#issuecomment-403527036


> I like the idea of adding API documentation, but mostly only if we can
> generate that from inline docstrings, as I'm not confident at this point
> that we can keep completely separate documentation pages up to date. I
> recognize we might need a page or two of overview documentation that would
> be separately written, but otherwise I think we can just use tools to
> generate docs and then make a habit of keeping the parameter explanations
> up to date in the code itself.
>

+1

There's quite good support for docstrings in Python, and well-establish
tool chains for publishing docs generated from them to the web.
The principle benefit of publishing docs on the web is that they can be
found by search engines.

I haven't ran the script yet, but there are utilities for finding docstring
coverage. Issue created here:
https://github.com/datactive/bigbang/issues/341

I'm not at all clear on what would be necessary for the Python 2 -> 3
> migration or what benefits that would bring us, so I hope you can take the
> lead on that Seb, if indeed it's important for future progress.
>

I can take the lead on this.

It's something we need to do to keep up with the dependent packages.
Not all of them will support Python 2 in the long term.
It will also make it easier to install, since most new Python installations
will be Python 3 going forward.


> What do others think? If this makes sense, should we start opening issues
> to track these tasks?
>

Already on it! Milestones sorted out here:
https://github.com/datactive/bigbang/milestones

Now the question is how to get the work done.


> On Jul 3, 2018, at 7:02 AM, Sebastian Benthall <sbenthall at gmail.com>
> wrote:
>
> Hello!
>
> Feeling fresh from the 0.2.0 release, I'm thinking about how to keep
> momentum. It was, really shockingly, over three years between the first two
> releases, and that's really not right. Research [1] has shown that after
> three releases, a project is much more likely to be a 'success', not
> getting abandoned.
>
> There is also clearly a lot of ways to polish BigBang that are not deeply
> technical.
>  - We made a lot of progress on the notebooks in the last release, but
> there is still lots more to do. For example, there's no reason why we
> shouldn't have notebooks demonstrating how to answer each of Corinne's
> questions from her recent thread.
>  - There's also lots we could do to improve documentation. We should be
> publishing the API docs to a website like https://readthedocs.org/
>  - We have a few automated tests, but not thorough test coverage. We could
> improve that.
>
> I propose we make this kind of polishing work the goal of the next,
> *0.2.1* release. This would be a small patch release on the existing one,
> with no major functional changes.
>
> A reason why I'm proposing this is that in my mind, the most urgent big
> update needed to BigBang is conversion from Python 2 to Python 3. That will
> involve a lot of tweaks across the entire system. Automated test coverage
> and good documentation of the existing functionality is important to make
> sure we don't lose quality and introduce new bugs when making that upgrade.
>
> What say you?
>
> Thanks for reading,
> Seb
>
> [1] https://mitpress.mit.edu/books/internet-success
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ghserv.net/pipermail/bigbang-dev/attachments/20180709/9811d5bc/attachment.html>