launchpad-dev team mailing list archive

[Robert Collins <robert.collins@xxxxxxxxxxxxx>]

So the lowest barrier to entry for 'looks like LP but really
different' that I can think of is 'ec2 demo' in the LP tree: this will
get you something with all the style and templates etc, which you can
have put plain ol html and js stuff on top of to do
whatever-crazy-you-want. 2 lines of ZCML and you have a fresh view, no
DB persistence needed - and you can do such persistence if you want.

prototypes is one of the prime use cases for 'ec2 demo', which is
unfortunately bitrotten, but thats easily fixed.

I think that by the time you want persistence and templating you're
really into the same environment LP *should* be efficient at
developing in.

That said, prototypes are a very valuable way to find issues as long
as we really aim to fail in an *informative* fashion: the prototype
should show us whats wrong in as many ways as possible, so that the
real version doesn't have those problems. If its more efficient to do
those as microsites in flask, django or pyramid, its fine by me.

I see one warning bulb here though: if we build a 'prototype server'
and invest in it - well, to me, thats worse than ec2 demo, because you
have cross team friction (different needs to meet), have to keep
updating it for style sheets etc.

There is an alternative though, which I think would rock, which is to
implement the template+API microservice from the SOA plan: that *by
its very nature* would be thin, lightweight, and easy to change : not
much different to a common prototype server, but with the significant
advantages that trunk for it *would be* trunk for the main LP. Not to
mention that making it easier to change LP is a big win itself.

-Rob