The concrete use case to be solved by this issue is to enable fast retrieval of change sets "linked" to a given set of JIRA issues.
We rely on this linking for release processes, and so enforce both the presence and format of the link. In our case the link is the JIRA issue key placed at the very beginning of the commit message.
Fisheye currently supports searching the full text index for a "word", or using a regex. Both of these methods can retrieve changesets based on a JIRA issue key in the commit message, but the first is too imprecise (picking up issue keys not at the beginning) and both are too slow. I would like to be able to retrieve the linked changesets for 100s of JIRA issues in a matter of seconds.
So to the proposal:
Fisheye could support adding user fields to the index, based on a regular expression. For example:
Then when indexing the repository, the comment (or source, or whatever) is searched for matches, and added to the index using the specified key.
This should then be available in EyeQL: