On 2014-06-03 11:52 AM, Brian DeRocher wrote:
What i do like about GitHub is how it facilitates forking and automated
testing. But i don't like how we are relying on the good will of a 3rd party
company.
As a maintainer of projects which use github, I understand the reservations.
Github has three parts to it
- Git hosting
- Forking and merge handling
- Issue tracking
For git hosting, we are not relying on their good will at all - it is
fairly easy to go and push a git repo anywhere else, or to self-host.
There'd be a bit of pain with the transition, but this is true for any
host switch.
Forking and merge handling has a bit more lock-in to it, but these
things tend to be short-lived in nature, so it's not a huge issue. One
alternative workflow here is the email-based on the linux kernel uses,
or some other git hosting site.
Issue tracking is more complicated. I understand peoples objections
here. There's two reasons why I still like it. The first is it
integrates well into the code handling and merge handling. The second is
that I'm far far more productive with the Github issue tracker than with
others I've used. Sometimes its a bit overly-simple, but it's so much
easier to use.
As an example, my latest issue/pull request is
https://github.com/gravitystorm/openstreetmap-carto/pull/599. If I had
been doing this in any tracker I've used previously and have experience
with, I'd of spent a lot of time wrangling images, formatting, code
references and copy and paste.
I use drag and drop images, cross-issue links, issue references from
commits, cross-repo links, references to commits via sha values, all the
formatting available.
As for automated testing, I don't use it. I know the capabilities are
there, but what I've been working on doesn't lend itself to simple
automated testing.
Whenever I weigh the positives and negatives of using Github, I always
come out in favor of using it.
_______________________________________________
dev mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/dev
