darkdreamingdan

The future of MTA Source control management: Your opinion.

Would you like to see MTA move back to SVN?  

45 members have voted

  1. 1. Would you like to see MTA move back to SVN?

    • Yes
      32
    • No
      4
    • I don't mind
      9


Recommended Posts

Over the last few weeks we've been discussing a solution to a dilemma that's been obstructing us for some time - source management.

Prior to the migration to GIT, we were facing problems involving the corruption of the main respository due to untested changes, and decided to switch to GIT's extremely flexible source control.

Unfortunately, GIT has proven to be no longer sustainable. Although it offers all the features needed for a diverse open-source project such as MTASA, it has slowly become impractical for the source mantainers to merge your changes into the source. This means although it's (relatively) simple for anyone to create their own forks and push changes, it's a painful process to merge this into the main MTA repository itself. On top of this, i'm sure many of you will be wary of the general complexities involved in setting up your own fork, and getting your head round GIT altogether.

We have now concluded that the best course of action will be to go back to SVN (and Google Code). However, a lot has been learnt from GIT and things will not be how they were originally with SVN. We have decided that becoming a Project Contributor will be much more lax than it was previously: if you've submitted a few (2-3) worthy patches, you will be added and have commit access to our SVN. This means we expect a large number of people to be able to contribute - maintaining the public nature that GIT offers. I would like to stress that it will not be difficult to gain commit access. Furthermore, commits will be reviewable by anyone in the public where they can add their comments easily to each commit, and give a positive or negative rating.

Of course, to prevent source corruption we will add a number of measures. Firstly, we will be moderating the SVN repository to ensure that it is always compilable and playable. Poorly coded changes will be reverted, and changes that are good enough will be filtered off into an 'approved' branch. This means we will mantain a partial structure of a development and unstable developer builds as with GIT.

The question i pose to you is: Would you like to see MTA move back to SVN? Feel free to be perfectly honest, and ideally state your reasoning in a reply to this thread. I must emphasise that although the verdict of the poll will influence the decision, it will not be the deciding factor. That is something left to the team as a whole. Feel free to vote even if aren't a coder - if you have an insight into the development process and understand some of the logic behind this decision, please do vote. If you have no idea what this topic is about, I'd kindly ask you refrain from voting.

Thanks

Share this post


Link to post

In my opinion, moving from SVN to Git was like moving from coding for hobby to coding for work.

I vote for SVN.

Share this post


Link to post

I vote for SVN, but i like the github style and the forks.

Googlecode website looks soo old :(

Edited by Guest

Share this post


Link to post

Same here, but how to filter the approved Snippets ?

Theres going to be a List like the Roadmap with approved Functions

and how the ppl. differentiate useful/non useful functions ?

In my opinion the Project gonna be messed up by Fools uploading

10-20 edited Header / Source Files for joining the SVN.

But i like the Idea uploading my patches into a respository where the ppl

can check out the Functions.

I gonna miss github :( Google Code is behind the times

Voted Yes

Edited by Guest

Share this post


Link to post

I don't mind - GiT was really hard to setup until I used SmartGiT. GiT is really bloated, but it's helpful for new developers. It got individuals coding, but kind of stifled MTA Development progress. :\ I'm kind of stuck on this one.

Ryden also had a point, coding in GiT was like coding for work. It felt so... corporate, not that it's a bad thing.

Share this post


Link to post

I think moving back would be a good decision. Sure that fork system makes it quite easy for patch contributers to get their changes applied and reviewed but it shouldn't be that difficult to get this working in a similar way with SVN.

As already mentioned a major downside of GIT is it's complexity. It's just way too easy to break your local repository and usually I had to delete and re-clone my complete fork in order to fix it up again. Seeing your patch lying around in the bugtracker for weeks is less frustrating than dealing with GIT in my opinion.

Share this post


Link to post
Same here, but how to filter the approved Snippets ?

Theres going to be a List like the Roadmap with approved Functions

and how the ppl. differentiate useful/non useful functions ?

In my opinion the Project gonna be messed up by Fools uploading

10-20 edited Header / Source Files for joining the SVN.

But i like the Idea uploading my patches into a respository where the ppl

can check out the Functions.

I gonna miss github :( Google Code is behind the times

Voted Yes

It's worth bearing in mind that we will revert poor commits, and those who are more frequent offenders will simply be demoted to having to submit patches again. As for the filter, we do have a system in mind that will work. There will be a seperate branch that the public are unable to commit to but will have selected changes picked by us filtered into it. This branch will have it's own changelog, so you'll be able to see the progress of the "real" MTA, and contributions by other people in another area. With these measures in place, in theory, the SVN should not get corrupted. If it does get completely messed up, it can always be synced up to our "real" MTA branch where things will always be stable.

Share this post


Link to post
My opinion: SVN with mirroring to GIT.

The trouble is, we do not want to mantain contributions to MTA via two different control systems. Even if we mirror changes to GIT, any contributions via GIT will be rejected. It's too much work to have to moderate two seperate sets of contributions.

Share this post


Link to post
My opinion: SVN with mirroring to GIT.

The trouble is, we do not want to mantain contributions to MTA via two different control systems. Even if we mirror changes to GIT, any contributions via GIT will be rejected. It's too much work to have to moderate two seperate sets of contributions.

i know, but GIT developers can provide modified patches for SVN.

Share this post


Link to post
In my opinion, moving from SVN to Git was like moving from coding for hobby to coding for work.

I vote for SVN.

I agree entirely with this statement really

source corruption measures look good

Please bring back Picard he was useful :)

Share this post


Link to post

Have other alternatives been looked at?

Share this post


Link to post
Have other alternatives been looked at?

Yes. Several, including hacks of SVN, other Git hosts, Bazaar, Mercurial..... None of them are ideal, and this is the conclusion we came to.

Share this post


Link to post

Anything that reduces the barrier of entry is good. I was one of the proponents of git in the first place - purely on principle, rather than after actually having tried it. The use of git was always on meant to be on a trial basis, and I think it seemed obvious to me fairly quickly that there are some fairly major issues with it (mostly in it's usability). I think going back to SVN is fine - it's not the 'cool' thing to do. Everything I've read has suggested that git is much better at merging branches than SVN is - I've little experience with doing that in either, but if that's been taken into account, I'd go for it. Just make sure there are systems in-place to make the processes smooth, and people to keep those systems running.

If only Accurev was open source!

Share this post


Link to post
To be honest things were 10x easier with GoogleCode. Voted back SVN.

Share this post


Link to post

We are moved to googlecode, topic lock.

Share this post


Link to post
Guest
This topic is now closed to further replies.

  • Recently Browsing   0 members

    No registered users viewing this page.