On 2013-02-08, at 21:19, Frank Shearar <frank.shearar@gmail.com> wrote:
On 8 February 2013 19:51, Camillo Bruni <camillobruni@gmail.com> wrote:
With the recent announcement of google code to shut down their public API I see a major functionality gone for our project.
https://code.google.com/p/support/wiki/IssueTrackerAPI
I would like to extend the functionality of our monkey, so it will become more intelligent: - give code critics feedback - reject new code that doesn't meet our criteria in general (unclassified methods / uncommented classes) - failing tests are serialized and attached to the issue - image with the changes integrated are attached to the issue
... you get the picture. All this stuff is impossible to achieve if there is no scriptable API available. By dropping that, google code becomes a silly toy with no further use to me.
Requirements ------------ So, we have to come up with a new issue tracker by june with the following requirements:
- dead simple issue reporting (most of the stuff out there just looks like a control panel for a space ship) - scriptable API - file attachements
Additionally: - programmed in ruby or python - easily create sub-projects ...?
Issue Tracker ------------- - I like trac a lot, but no API from what I read http://trac.edgewall.org/ - Jira, thats the space-ship-panel (and in this very same category, buzilla...) http://www.atlassian.com/software/jira - github? too simplistic no file upload https://github.com/dalehenrich/filetree/issues
What's too simplistic, in particular?
There's no file upload, but it does have an API and you could post URLs to some server somewhere. What would be really cool is using the git tools: you have a slice, you branch off master, apply the slice, push the branch, and get github's diff.
ok, that would work for most issues, but how do we deal with changes that require pre and post-scripts? Until we can rebuild our full image from sources we need to have pre- and postscripts for certain critical issues. Currently these scripts are communicated to the integrator, which is a very bad thing. I want to have that done automatically. => we need file uploads for a while we might be able to link gists into an issue. But I would prefer real files for now. June isn't that far away and I don't want to change too many variables at the same time.
You can signal bad pull requests by setting commit statuses on commits: that API's pretty straightforward (http://developer.github.com/v3/repos/statuses/). So the monkey could apply lint rules etc., and colour the commit accordingly.
I know of all of that and it is certainly what I want in the future but: - we do not run on git yet - we replace our main issue tracker so I would like to have something that works as close as possible to the existing solution we have.