Closed budu closed 13 years ago
budu
commented
13 years ago Also, I don't understand what have been done with branches. I know there's plenty of ways to organize versioning with Git, but the way I'm using it is to make a branch for a "major.minor" version only if it divert from the master branch, then add fixes to those branches and mark them with tags. The way the "1.0.1" has been created, it's really not quite clear what "hotfixes" has been included.
doooks
commented
13 years ago I don't really care if people want to use branches in a unconventional way, but releases have got to be uniquely version numbered, and the repo should really be tagged with a matching number so builds can be mapped to commits.
If normal release conventions are followed I don't have to waste time working out if I have a bug fix in my jar or not. I just look at the version number.
LauJensen
commented
13 years ago Budu, where have you seen the branch being rewritten? I remember doing this frequently in the first few days, simply because I assumed nobody was watching - I agree this is bad practice!
The idea of hotfixing a version is not something I have invented, though I know its not standard in the OS communities. It is also something I hope to avoid in the future, though no guarantees.
In the future I hope to use the new Cake versioning features, which means that me and all contributors can tag commits to bump versions on bugfixes etc. I will make a formal announce about this before it takes off.
budu
commented
13 years ago Budu, where have you seen the branch being rewritten?
There was one a little more than a week ago, on Friday the 7h, r0man accepted a pull request (Issue #56) from me including a small fix to make it possible to run the tests with Leiningen. It disappeared during last week (I don't know exactly when) breaking all my local branches.
In the future I hope to use the new Cake versioning features, which means that me and all contributors can tag commits to bump versions on bugfixes etc. I will make a formal announce about this before it takes off.
Than would be great, I've not looked at that cake feature yet, I look really awesome as you described it!
LauJensen
commented
13 years ago Ah - I remember. I dont think that MASTER was rewritten. Roman had pulled a patch which I hadn't approved, so I branched, rebased to exclude the patch and merged again. As far as Im concerned thats common practice. Im sorry that it caused problems on your end though!
Regarding the new cake features, check this out: https://github.com/ninjudd/cake/wiki/Project-Versioning
budu
commented
13 years ago so I branched, rebased to exclude the patch and merged again
That amount to rewriting commit history, as soon as you use rebase to remove a commit that was already pushed that's what happen. The correct way to remove that patch would have been to actually remove the unwanted code and commit the changes. Maybe there's another way I don't know of, but that's what I'm usually doing when that situation happen. Also, git prevent pushing rewritten history, you certainly must have pushed with the -f option at same point.
LauJensen
commented
13 years ago I see what you're saying and you may be right - Sorry about that. Im glad you raised these issues, I will certainly refrain from such rewrites in the future.
LauJensen
commented
13 years ago Accepted and closed
I've got two issues about project management:
Firstly, the way the public repository is being handled. Today, I've seen that the commit history of the master branch has been rewritten. This really shouldn't happen and I don't see any reason why it happened. Can anybody comment on what happened?
Secondly, semantic versioning. There have been a "hotfix" of version 1.0.0, then again, I don't understand why this has been done, apart from (bad) laziness, it should have been a separate release.
One thing I really like about the Clojure community is that most projects seems to respect those fundamental aspects of project management. I'd like to see the same from ClojureQL.