sebb wrote: > On 24/01/2010, Phil Steitz <phil.ste...@gmail.com> wrote: >> sebb wrote: >> > release-process.txt currently says: >> > >> > 7. Check out DBCP fully - i.e., trunk, branches, tags. >> > >> > 8. From the top level of the checkout, execute >> > svn cp trunk branches/DBCP_1_3_x_BRANCH >> > svn commit -m "Created 1.3.x compatibility release branch." >> > (where x is the final digit of the new release number) >> > <<<< >> > >> > This seems like a lot of SVN download traffic, and the amount will >> > increase with every new branch or tag added to DBCP. >> > >> > Surely one can just create the tag directly in SVN, e.g. >> > >> > svn copy -m"Create 1.3.x compatibility release branch." \ >> > https://svn.apache.org/repos/asf/commons/proper/dbcp/trunk \ >> > >> https://svn.apache.org/repos/asf/commons/proper/dbcp/branches/DBCP_1_3_x_BRANCH >> > >> > The newly created branch can then be checked out locally. >> > >> > Or am I missing something here? >> > >> > We could create a script for creating the branch, and probably quite a >> > bit of the rest as well. >> >> >> I use a script for most of the steps. Not everyone uses bash >> locally, though, so I did not commit it. > > Still worth committing; I can convert it to a Windows CMD file. > Between the two, that should be enough. > I will either commit or embed the commands in the process doc. I do not script the whole process end-to-end, as there are verification steps included.
> Or maybe it can be done in Ant, in which case it's cross-platform... > >> The download traffic is essentially the same when you create the >> branch server side or locally, since you still need to download it. > > Only if you want the rest of the SVN branches and tags, which I don't. > >> I personally prefer to do things locally and then commit. >> Admittedly, this is just a matter of taste and I am fine with >> changing the instructions to use server side copy. One thing missing >> from the instructions is that you need to rm any previous branch >> created with the same name. >> >> Step 7 obviously only has to be performed once. I personally >> maintain full checkouts of all of the commons components that I work >> on. I understand that others may think this is excessive; but it is >> convenient to have at least all of the release tags checked out, and >> as I said, it only has to be done completely once and then updated >> incrementally. If this seems to pig-ish, I guess I could revert to >> manually grabbing just the release tags and branches. Right now, a >> full checkout of [math] (the biggest one I have) is 501MB. DBCP is >> 211MB. > > It's up to you what working copies you maintain; I personally would > not want all of those. > > So I propose adding the alternative method of creating the branch for > those prefer a differnt way of working. OK? That is fine by me. > >> Phil >> >> >> > >> > --------------------------------------------------------------------- >> > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> > For additional commands, e-mail: dev-h...@commons.apache.org >> > >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> For additional commands, e-mail: dev-h...@commons.apache.org >> >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
