Technologies

[Tom MacWright]

Downvote to HAML. It's totally unknown outside of ruby-land and would
significantly cut down on the number of possible contributors to the theme
layer, nixing the two or three people I know who are working on patches.

Downvote to selenium; MapBox used it on TileMill and it was a headache in
every way - setup, maintenance, false-positives (and negatives). The
interface level of OSM is not large enough nor complex enough to warrant
that kind of testing - it's more important to get the Rails-level tests
working well.

On Fri, Aug 31, 2012 at 9:28 AM, Shaun McDonald
<shaun at shaunmcdonald.me.uk>wrote:

>
> On 31 Aug 2012, at 10:31, Andy Allan <gravitystorm at gmail.com> wrote:
>
> > On 30 August 2012 22:49, Tom Hughes <tom at compton.nu> wrote:
> >> I commented in one of the pull requests I merged a few days ago that I
> had
> >> chosen not to merge a commit that added some tests because it introduced
> >> some new testing technology and I thought we should discuss that before
> >> making a decision.
> >
> > It's a long time since I've looked at the rails-port tests. There's a
> > lot of things I'd like to change, but I'm wary about changing for
> > change's sake - given that I know how long it took Shaun to write most
> > of them originally, and that I'm unlikely to be spending a similar
> > amount of time to re-implement things! So take what I say with a pinch
> > of salt.
> >
> >> The specific technology in question was the capybara testing framework
> along
> >> with the selenium driver.
> >>
> >> Personally I don't see any problem with capybara, though maybe some of
> the
> >> other rails people (Shaun? Andy?) here have preferences for something
> else?
> >
> > I use capybara for testing pages, I don't have any issues with that. I
> > don't, however, use any of the javascript drivers, so I can't comment
> > on selenium vs webkit. When I've tried getting selenium working
> > before, it's been a bit of a nightmare.
>
> selenium is quite difficult to get working, though does give the advantage
> of being able to test the js on many browsers, though I'm not sure how best
> to test that the maps have loaded correctly.
>
> Webkit is more reliable in getting working, though a year or two ago I
> came across and issue where someone in the team couldn't get it to work.
>
> The browser based testing technology is quite fast moving, though I think
> the technology choices are starting to settle down compared to a couple of
> years ago.
>
> >
> >> Anyway, my more general question is what thoughts any of the rails
> experts
> >> here have on what new technologies we should be considering employing if
> >> any...
> >>
> >> That includes both things like testing tools, which I know Shaun has
> >> expressed opinions about in the past, and things like alternative
> template
> >> languages - I think Andy has suggested HAML in the past for example.
> >
> > Yep, in summary:
> >
> > * FactoryGirl as a replacement for fixtures
> +1 this will also speed the tests up.
>
> > * shoulda-matchers
> No opinion
>
> > * rspec instead of test::unit, but I doubt it's actually worth the
> hassle.
> Doesn't make much difference. It's just a different way of doing things.
> test::unit is supposed to be slightly more efficient as it doesn't use the
> missing_method stuff. I wouldn't say it's worth the change.
>
> > * haml instead of erb
> +1
>
> > * devise
> > * declarative_authorization instead of scattered authorisation around
> > controller actions
> +1 it does take a bit to get your head around that way of doing things,
> however makes things much more consistent.
>
> Shaun
>
> >
> > Cheers,
> > Andy
> >
> > _______________________________________________
> > rails-dev mailing list
> > rails-dev at openstreetmap.org
> > http://lists.openstreetmap.org/listinfo/rails-dev
>
>
> _______________________________________________
> rails-dev mailing list
> rails-dev at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/rails-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/rails-dev/attachments/20120831/4c029583/attachment.html>