[OpenStreetMap] #4246: gpx_import fails on a ruby error (Magick)

Andy Allan gravitystorm at gmail.com
Thu Mar 1 19:38:36 GMT 2012


On 26 February 2012 14:19, Tom Hughes <tom at compton.nu> wrote:

> But we're not talking about theming here anyway - we're talking about how
> far we go in maintaining code that is no longer used by the OSM web site.

Well, we're talking about the slightly wider point about whether we're
building some software that has one, and only one, intended use. Or,
as an alternative, are we building software that has an (obvious)
primary use, but is also useful for other people.

You know as well as I do that other people install the rails port
because, for a wide variety of reasons, the data they want to work
with isn't suitable for inclusion in OSM. Take Peter's 1930-whenever
railway map, to give just one example. So there's a demand for
works-just-like-OSM-but-isn't-osm.org websites.

> Should we have kept MySQL support working after switching to Postgres in
> case somebody preferred that?

I put in an enormous amount of effort to make sure that all the
migrations worked on both systems, back when we changed over. I've no
idea whether anyone was taking advantage of it, but it's now it's
impossible because you yanked it.
beb0ef6357e426ab08349109d47c3726ed50042e

> Should we keep the ruby GPX importer working when we don't use it? Even if
> that makes it harder for new developers to get up and running?

You want to yank the GPX importer mainly because of the RMagick
dependency - but we could remove that small feature and leave the rest
intact.

> Should we keep all the API code as we gradually migrate the OSM site over to
> using cgimap to handle the API calls?

... and so on and so on. There are endless examples, and clearly
there's a balance to be struck.

Your point appears to be that if it's not used on OSM.org, you're not
interested in it being in the codebase at all, regardless of whether
anyone else finds it useful. It's an extreme position, I'd say. Given
you're the project maintainer, I urge you to change your mind! We talk
all the time about how few people we have working on the code, but by
repeatedly making it *less* useful to other people and other projects,
I don't find it very surprising that there are fewer developers.

If we head even further towards making the code only useful for
osm.org, then it becomes even more important to grab core
functionality - e.g. the rails implementation of the 0.6 API, clearly
useful code - and separate it off so that it can have a life of its
own.

Cheers,
Andy



More information about the rails-dev mailing list