> > > Here's what I found: > > 1) It looks like you were able to preserve the history in the > migration from Hg to SVN. Good job! What tool did you use for this? > > Use the Shell script I mentioned before with some small updates for our purpose. And run it on WINXP in Cygwin.
2) I received a checksum error when unzipping the archive. I tried > this several times, downloading the zip file on two machines. I > always got a checksum error on the simple api dump. > > 3) I also got a checksum error while processing the load of the odfdom jar. > > Strange... I test it on my machine, there is no error... 4) Other dump files, like the conformance tools, processed without error. > > 5) I was not able to load these all into the same repository because > there are conflicting files, e.g., multiple projects have a README or > license.txt in the root. > > I'm not sure what is causing the checksum errors. Could you try an > "svnadmin load" for odfdom and simple and see if you get these errors > as well. If not, then maybe you need to re-zip them and upload them > again. > > I try "svnadmin load" on my machine, it works well. I even checked out them, all of them without problem. I didn't take my work laptop back home, so have to re-zip them and upload tomorrow(night in your time). Also, to get this all into a single repository we need to agree on a > directory structure. When we do an svn adminload we can specify a > parent directory (-parent-dir) , so we can do that to work around the > conflicts. > > One solution is to load each dumpfile into its own directory, like: > > /odf/trunk/generator > /odf/trunk/odfdom > /odf/trunk/simple > /odf/trunk/old-generator > /odf/trunk/taglets > /odf/trunk/validator > /odf/trunk/xslt-runner > /odf/trunk/generator-task > > For example: > > svadmin -parent-dir /odf/trunk/generator ./odf-toolkit-repos < > generator_svn_dump > > svadmin -parent-dir /odf/trunk/odfdom ./odf-toolkit-repos < > odfdom_svn_dump > > and so on. > > Then when we have one big combined load repository, we can create one > big dumpfile for it, and that would be what we sent to Apache Infra > team. > > That is the "quick and dirty" approach. Once imported, and after some > more thinking, we will likely move things around, create a top-level > build script for the entire toolkit, etc. > > Comments? > > Hehe, no comment;) That's reasonable. I will create a new single big dump and upload. -Rob -- -Devin
