Now, there are three questions we have to decide:
1) About the repository structure. Here, we should keep in mind, that: Every repository has its own history. Branching and tagging is always repository-wide. When you initially want to "download" the source code (git clone), only complete repositories can be downloaded. Therefore, I propose to have exactly one repository with the structure as it currently is in http://git.softeys.de/tmf.xtext.git/tree However, I noticed that our releng-stuff is in
http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.tmf/org.eclipse.xtext.releng/?root=Modeling_Project
rather than
http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.tmf/org.eclipse.xtext/releng/?root=Modeling_Project
Is there a specific reason for having this special top-level-folder? Having it in the sub-folder "org.eclipse.xtext/releng" would simplify and unify the structure.
That is because during the build we need a posibility to check out the releng project only, without hundreds megabytes of
"unnecessary" artifacts. It is also important that releng project don't share the same directory with the rest of bundles (simpler scripts easy cleanup etc.)..
2) Import history vs. import import HEAD only. I propose to import HEAD only, since with each "git clone" one has to download the repositories complete history which for us is currently 170MB. We should have a second, read-only repository containing the complete history for the cases we need to do some archeology. I've asked the webmaster whether this is possible: https://bugs.eclipse.org/bugs/show_bug.cgi?id=317923
Du meinst, in 2-3 Jahren haben wir unsere
Geschichte noch einmal deutlich, denn git-Repository wird zu groß? Mit CVS oder SVN
auschecken wir nur den aktuellen Stand und die Geschichte kann bei
Bedarf abgerufen werden. Ich denke also, mit git es sollte auch möglich
sein, zu konfigurieren, wie viel Geschichte meiner Klon enthalten wird.
Regards,
Dennis.
