Hi Buggy! Am Samstag, 14. März 2009 01:30:39 schrieb bugs buggy: > Sounds ok, I am just not sure how we can shift the load around, since > we don't really have enough people to shift the load, and the same > people end up doing it all the time, and when those people are away, > then we are sorta stuck, until someone else picks up the slack. Giel was away. Per and I picked it up and released 2.1.2. Seems to work.
And if we decide we are lazy and do not need QA, we can still simply merge everything that moves. But it seemed to me that it was desired to add a bit more QA, have patch reviews established and all that. It would come naturally if the primary repository was to be "guarded" by someone. I was told that it was not desired to have a version in VCS which does not compile or run, introduces new bugs, etc. If you make the releases (and their branches) guarded by someone who actually does QA, has the time to fetch changes from everyone every once in a while, then you get what was asked for. If you decide not to do any extra QA and just fetch&merge, then you get what you have already at not a lot of extra cost. > Perhaps, doing it the 'git' way is a bit more complex to understand > how to organize everything. Yes, it involves a bit getting used to. And it has a different form of organisation. It involves a little bit more communication, which is imo not a bad thing. Prevents everyone from doing their thing and just throwing it at SVN. What I expect is more review of commits, and possibly even of patches. > AFAIK, we all need to make accounts with gitorious, then we do a clone > to grab everything. Do we need to be assigned/attached to the > project first, or can we do that after the clone? No, you do not need to signup at gitorious. All that you need is some way to make your work accessible to others. How you do it is entirely your choice. If you want to offer ssh accounts on your home machine? Fine! You have a webserver setup on your machine and it is always connected to the net? Fine! You have some spare webspace somewhere? Fine! You want to setup a git-daemon on your server? Fine! You send patches via Email? Fine! You send patches via Instant Messenger? Fine! You host your repository at GitHub? Fine! You want to host it at Gitorious? Fine! ... > Once that is pulled down, we have our own copy of whatever is on > gitorious, just like if we were to do a svn checkout. No, with svn you only have the last revision. Want to do svn-diff on non-head? Do it on the net! Want do switch the checked out revision? Do it on the net! Want to look at the log? Do it on the net! Want to svn-blame someone? Do it on the net! ... Needless to say this annoys me a little, especially if you are on a slow or unstable connection... > Now, if we pulled down everything, do we do: > git branch <- lists all branches > git branch 2.2 <--I want to work in this one I think you want "git checkout --track 2.2" here, but I am not exactly sure. > git checkout 2.2 <- set it to this one. Note that it is probably more like origin/2.2, which you checkout. (Or devurandom/2.2, buginator/2.2, etc.) > now, how do we work with the files, since they are still in the main > directory. I am not exactly sure what you are explaining here... > I mean, I know how git handles it, but how do external programs > handle this? Which external programs do you refer to? And which issue might they have? > It looks like we must do a git clone some\new\directory, from > the initial clone directory, then work in some\new\directory, and when > your are done with that, you push the changes to the initial cloned > stuff, and then push upstream correct? Actually I checkout whatever branch I want, work on it, maybe look at other branches (but I dont check them out), perhaps stash my changes to switch to a different branch or do some intermediate commits, maybe do some temporary commit to save my work or switch to another branch to continue my work on another day, etc. > Lastly, about .gitignore. I assume we should add all the windows > specific stuff to that ( *.obj, *.lib, *.exe, *.htm, *.pdb, *idb, > *.ilk, *.res, *.manifest, most likely some others as well)... Feel free to add it ot a new section in that file. (But please keep it ordered and commented.) Some sidenote: Maybe you noticed it already: Even if "Gitorious" (or whatever is your host) suddenly goes broke, is shut of by your govn, becomes victim of a fire, or decides they dislike you: You can continue to work, commit, etc. You will have to find a different way of publishing your changes, yes. But there are enough channels of communication nowadays, that this should really be no issue. Greetings, DevU P.S. If Git does not feel like the ultimate solution to you, and you want to propose something else, I wont object against a test. I have been using Mtn for a while, which was quite good. Bzr was not bad either (it felt slow, but that was ages ago, maybe it is better now). I have been consuming Hg several times, too, and it seems to spread quickly. I also used Darcs, but it is said to be slow with large repos, and if you are not on the commandline, it might be difficult to use (unless there are GUIs). With none of them I have as much experience as with Git, but at least the basic tasks (clone, update/pull, commit) work in all of them quite similar and convenient. We would have to test how they compete with Git when it comes to the more complex tasks, like publishing changes (push/format-patch/send- email/...), merging, and so on.
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
