No, we can’t split the information.
A “partial switch” could be misleading. You can read from the above comments that the testdev1 is a source of confusion by now . Of course, we can always change idea and go back to Redmine in the future, but I’m pretty sure that if we do the change we’ll be happy with it.
I agree with @Alessandro_Lepore and @alefattorini here: we should encourage new contributions at all.
Moreover, I think from a developer’s perspective having both issues and pull requests together on the same platform is a plus.
valuable point of view
yes, I spoke of that some month ago
you are not objective
yes, you must be able to sort issues following the NS version target or whatever you are looking for, bugzilla does it, like any ‘good’ software of project management.
I tend to agree, I recall the conversation I have had with a friend concerning github and the community around it, you can attract developers, much more than at sourceforge.
Well hard to say, the multi repository is really a bad point
I don’t understand why we are still talking about multiple repositories. The issue tracker would be into just one, dedicated, repository. This is a common pattern on GitHub hosted projects.
Indeed I missed the train, I’m playing a bit with github, fail2ban can wait
i was just going to ask this … could you explain the pros and cons of having a single repo for the issues or have the issues for each repo? (sorry not a long-time github user here )
for what I understand, If we use a single repo, we must use additional labels and leave the issues “disconnected” from its repo. What happens then with the PR?
do you have some examples of projects using this solution?
if we decide to use the classical approach (each repo has their own issues) we can use the search functions of github to see the issue of the entire organization … or could we try zenhub? It seems free for open source project and it support multi repo
for the rest I agree with @stephdl
@giacomo I apologize right now but I will do other tests with testdev1… sorry for the email
No chance to have page with a full witdh on github, it is really a pain…Santa is bringing me three 75’’ for christmas
More seriously the search query is better in redmine, you cannot have a fine search in github, or it becomes really complicated.
At the contrary it seems to me that the search issue is very very powerful on github but you have to know the correct sintax
Nobody wants full witdh anymore!!
when you need help, you missed something good in the design…
-
Issues on a single repo are a central point to collect bug reports and new feature requests.
An issue could be related to more than one package/repository. This is the reason for “multiple-packages” category in our Redmine. With GitHub labels it’s easy to assign an issue to multiple packages. -
Issues on a single repo have an unique numbering.
When the developer writes a commit message or a comment from another repository he can reference the issue by using a syntax like:nethserver/dev#1234
.
This is an example of pull request that references an issue of another project:
- rkt fails to render Docker images with whiteout-ed hard links · Issue #1653 · rkt/rkt · GitHub
- handle hard links and whiteouts correctly (v3) by alban · Pull Request #105 · appc/docker2aci · GitHub
A link between the issue and the PR is created with the syntax above, in the issue title, or comment. Consider also the scenarios
- multiple PR from multiple repositories referencing the same issue
- a PR without issue
OwnCloud uses a “mixed” approach. Main components have their trackers, the “apps” share a single one. https://github.com/owncloud/core/blob/master/CONTRIBUTING.md
I’d prefer not relying on other tools. The waffle.io dashboard is for @giacomo that loves summaries
However the Milestone view is a good starting point to catch the state of the whole project, together with the organization dashboard. Also queries can be saved somewhere (wiki.nethserver.org?) or bookmarked: NethServer v6.7.
What do you mean with “fine search”? What are you searching for?
agreed, one example Bugzilla Reports - SME Server
actually you can search easily by milestone or labels, but in redmine you have much more settings, however like @alefattorini proposed a search syntax exists
It’s a common approach, Discourse use a similar sintax
https://meta.discourse.org/t/advanced-search/2267/13
I admit Redmine has an advanced and complete search form, but I must also say I rarely use it. And I’m a top user of our Redmine
Checkin last two months activity you’re a top user togheter with @giacomo @filippo_carletti @jobezic @alep @dz00te @Adam
http://dev.nethserver.org/projects/nethserver/activity
I’d like to propose a real world test: use github instead of redmine for the “initial work” on NethServer 7.
Initial work means some small adjustments to a number of packages to make them work on CentOS 7.
My idea for a workflow is:
- clone nethserver-xxx source
- try to build rpm
- fix build errors (if present)
- install rpm
- create pull request
- open issue for unsolved errors
What do you think?
Okay, let’s give it a try for NethServer 7 since we do not expect to have many major issues during the porting.
If the GitHub tracker is not good enough, we will switch back to Redmine. In this case, probably we can also afford to lose few more issues (or we can copy them by hand).
At the end of the porting work, let’s say after the release of NethServer 7 beta, we will gather all the opinions to get final decision.
I will check it, but i’m a redmine fan so my opinion will be a little sided
, it was a breakthrough tool for the company i’m working.