We currently host vim-Qt at the following locations:
We try keep all these in sync, so that you can fetch from any of these services and trigger pull requests at any of these services. However issue reporting only happens here at Bitbucket.
General netiquette applies, just be civil and patient. Please include detailed information with the bug report, it helps narrowing down problems:
- Operating system/Linux distribution
- Qt version
- Sometimes some information about your vim configuration may be helpful too
- If possible chek if the vim-gtk version also has the same issue
As a rule of thumb the master branch should always build cleanly. This is also the branch from where packages are occasionally generated.
It is not uncommon to see topic branches - named as tb-somefeature - for more experimental features. Keep in mind that these may involve forced updates when a rebase occurs.
For bug fixes when some testing is desired - before going into the master branch - a fix branch may be created e.g. fix-cursorisbroken.
Periodically we merge with the upstream Vim repository(default) - https://vim.googlecode.com/hg/ - but we do not keep Vim's history on our repository. The merge event shows up as a merge with a very large commit holding all the changes. The rationale for this is that we should be able to easily rebase our changes on top of other vim trees.
In a nutshell merging vim-qt against vim happens as follows
1. Get the vim-qt source using git (git clone)
2. Create a new branch (e.g. git branch upstream) with the upstream version of Vim
$ git branch upstream $ git checkout upstream <erase old files> $ cp -r ../vim/upstream/* .
3. Commit the upstream version into the branch (git commit -a)
4. Create a new branch to stage a merge
git branch merge-upstream git checkout merge-upstream
5. Finally merge with the master branch
git merge master
6. Update the configure script (autoconf)
7. After resolving all conflicts and checking for errors, merge merge-upstream into the master branch
Step 2 looks a bit weird, it makes more sense if you consider you have a paralell branch to track upstream (which is what I do).
Here is a snapshot of the current infrastructure:
- git multipush is just an alias I have to push to all repos in one step
- For some reason launchpad will not import from smart-http(s) that is why it is pulling from github instead of bitbucket