I have an extremely large Subversion repository being indexed in FishEye. It's a few million lines of almost 20 year old code that is heavily branched. Since SVN branching doesn't do client-side diffs, rather they're full copies (at least for our version), it means the FishEye index is huge (~10GB) for this repository.
To make things wose, our FishEye instance is running on a virtualised platform with particularly slow disk I/O, so loading up a file that's been edited a bit over the years can take 10 minutes in some cases.
I've just read the diff info documentation and am now wondering if disabling diff info will help speed up this repository.
In what cases will the repository be faster? What sacrifices will we be making if I switch it over? e.g. can we still load up HEAD code quickly, even for branches? will we suffer worse performance (since it's a remote SVN over HTTPS server) on anything in particular?