[IPython-dev] bug in html notebook in latest version?

Matthias BUSSONNIER bussonniermatthias@gmail....
Tue Oct 23 12:56:52 CDT 2012

Le 23 oct. 2012 à 19:52, Brian Granger a écrit :

> On Tue, Oct 23, 2012 at 10:11 AM, Matthias BUSSONNIER
> <bussonniermatthias@gmail.com> wrote:
>> Le 23 oct. 2012 à 18:42, Brian Granger a écrit :
>>> On Tue, Oct 23, 2012 at 9:39 AM, Matthias BUSSONNIER
>>> <bussonniermatthias@gmail.com> wrote:
>>>>> I hate to say this, b/c I'm not the one writing much JS code and I
>>>>> don't want to sound like a grouch, but I'd like to ask everyone
>>>>> working on JS to tackle the testing problem as a very high priority
>>>>> question and to consider a moratorium on major new work until we sort
>>>>> it out.  For years the old IPython codebase started without tests, and
>>>>> that started to slowly paralize and kill the project.  Eventually we
>>>>> bit the bullet and developed a  testing machinery that, while ugly at
>>>>> the beginning, got us moving.  And over time, that machinery has
>>>>> gradually gotten better and better, and today on the core code we
>>>>> actually do have decent testing (not perfect, but not terrible
>>>>> either).
>>>>> I am growing increasingly worried about more and more JS code without
>>>>> tests and I fear we're going in the same direction here.
>>>>> I am NOT trying to slow down development or boss anyone around, but we
>>>>> really need to take this problem seriously, or the whole project will
>>>>> suffer.
>>>>> Thoughts?
>>>> I think our current JS architecture is not designed for good testing.
>>>> I'll will try to come up with an IPEP, but for the basic, we should:
>>>> make use of `require` and have a more MVC approach.
>>>> -> it will allow to have model testing on server-side with node for example.
>>> Do you have thoughts on the best implementation of "require"?
>> I've heard of requireJS,
>> and it ships with r.js that apparently allow to transform commonJS require to a more async requireJS form.
>> I've never used it though.
>> I think that the commonJS
>> x = require('…')
>> is much more readable and understandable than the callback based one.
>> In the long run, I think that having a 'dev' flag would be great.
>> Without the dev flag all the JS/css/coffescript/... would be minified and in one file.
>> and if you want to introspect, you activate the dev flag and everything is unmignified and readable.
>> RequireJs Does allow that, which is a plus, even if it means adding a compile option to js.
> OK thanks, I will try to have a look at these options.  Would  be nice
> to improve this situation.
I'm writting an IPEP. I should send the draft on github soon. 

> Cheers,
> Brian
>> --
>> Matthias
>> _______________________________________________
>> IPython-dev mailing list
>> IPython-dev@scipy.org
>> http://mail.scipy.org/mailman/listinfo/ipython-dev
> -- 
> Brian E. Granger
> Cal Poly State University, San Luis Obispo
> bgranger@calpoly.edu and ellisonbg@gmail.com
> _______________________________________________
> IPython-dev mailing list
> IPython-dev@scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-dev

More information about the IPython-dev mailing list