launchpad-dev team mailing list archive

[Deryck Hodge <deryck.hodge@xxxxxxxxxxxxx>]

Hi, Gary.

On Sun, Feb 13, 2011 at 9:02 PM, Gary Poster <gary.poster@xxxxxxxxxxxxx> wrote:
>
> I was hoping someone else with more knowledge than myself would step in to say this, but I'll jump in.
>

Sorry I wasn't able to weigh in before now.  I saw this over the
weekend, and was hoping to comment, but my weekend was just too crazy
with my daughters' various Valentine's parties and activities. :-)

> This situation is precisely why I currently advocate what Brad described.
>
> Here's a summary of my understanding.  Please correct.

[snip]

I'll follow up later and clarify a couple points related to your
assessment of the js packaging story, but generally this part:

> Therefore, as I assemble the picture around me, we have a broken JS packaging story now that is both significantly in flux and not necessarily appropriate for including all of the sorts of things we need in our local applications.
>

Is more or less dead on.  I would probably phrase it as "an incomplete
JS packaging story that is in flux", rather than calling it broken,
but your point is generally correct.  It's not always obvious where to
put js code, how to put there, and then how to build it once that is
done.  I'm trying very hard to work with Paul to make this better, to
share knowledge around this, and to eventually build some docs.  I
have some emails to send today about my current lazr-js work as well.
I didn't do it on Friday just to avoid this kind of weekend bikeshed.
Bikshedding is *always* better on a Monday and with fresh coffee, IMHO
of course. ;)

Also, Brad and I chatted and talked at least a couple times during his
development of this widget, so it's not as if he made his choice
without input.  I don't recall my specific advice, but I'm sure what I
said would have pointed to our current rules for where to put js code:

* YUI dependencies live in lazr-js
* New widgets should be developed in Launchpad first, then pulled back
to lazr-js when wider value is obvious.
* Gallery code is unique and we have to deal with it on a case by case
basis until we have a combo loader/Canonical CDN

So I think he made a good call here.  He got some work done he needed
to do.  He did it in a way that allows us to transition easily to
lazr-js if this widget works out.  He also went a step further and did
it in such a way that helps us pull the other modules like this out of
icing and into a sane location, where that code can then be assessed
for inclusion for lazr-js.

I support what Brad has done and think over-thinking it at this point
is premature.

Cheers,
deryck


-- 
Deryck Hodge
https://launchpad.net/~deryck
http://www.devurandom.org/