← Back to team overview

launchpad-dev team mailing list archive

Re: Do we need the lazr namespace in our javascript?

 

On Thu, Jun 21, 2012 at 10:16 AM, Curtis Hovey
<curtis.hovey@xxxxxxxxxxxxx> wrote:
> On 06/21/2012 10:11 AM, j.c.sackett wrote:
>> On 06/20/2012 01:42 AM, Huw Wilkins wrote:
>>
>>> When we pulled the external dependency into Launchpad we discussed that
>>> YUI and yui-gallery should be considered our upstreams. I still think
>>> this is a good idea, in that it would force us to write extensible,
>>> clean, documented widgets and it will mean that we'll be properly
>>> contributing back to the open source YUI community.
>
> We intended lazr.js to be our UI library that could be be used by all UI
> developers. We even dared to dream that as an early adopter of YUI3, our
> widgets would be the premier UI library. We adopted the documentation
> guidelines so that out work could be used by other communities. We were
> naive. We did not know how to build proper YUI widgets.
>
> /me is fixing the bug-tag widget that is not a widget.
>
>> I remember that conversation and still think this is a good idea; having
>> our own gallery-esque space didn't really work out--there's lots of
>> reasons for that, but the net take away is rather than silo-ing our work
>> if we want it shared we should share at the best level.
>>
>> That said, before we move too far in the "shared code" conversation, I
>> take it that there are no disagreements with removing the "lazr"
>> namespace within the LP javascript?
>
> There was a recent discussion to have a Canonical JS UI library. When
> designers update the library, they update all Canonical sites. As an
> organisation we want it. We do not know how to have such a library
> because the principle users have conflicting design interests or
> infrastructures that must interact with the library. lazr.js failed
> because u1, landscape, and launchpad have divergent interests.
>
> I image we will want to reuse our js code as we do more non-launchpad
> projects. We probably want to reduce Lp line count by creating a project
> that we control and package easily so that landing a feature does not
> require painful orchestration of landing and deploying two branches
> simultaneously. This would not be a lazr project. This would be
> launchpad-js, or something to do with our new name [1].
>

I think the problem with laza.js was that it was premature.  There was
nothing built in lazr.js that was reusable in any way.  Even now, we
only have a couple js widgets shared across lp, maas, and maza-server.
 So I think talk of setting up anything too formal is again premature.
 I suggested to Rick when he brought it up to me recently that we get
these couple shared widgets in the YUI gallery as a start.  If sharing
continues, then we could go deeper.

As Curtis notes, one problem is that we have so much disparity and
churn between all our web projects.  I don't think we gain anything
but trying to force less disparity or less churn.  I think it's better
to have anything that can be shared to be done in a form that makes
sense given the churn/disparity.  The YUI Gallery makes sense to me,
and also, it makes you really ask "Is this genuinely generally usable
by others, or just something we're passing around ad hoc in a couple
places?"

This works better for js widgets than css, IMHO.  I'm not sure what
the answer for shared CSS should be, or if we even need an answer
there.

Cheers,
deryck


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


References