We'd really like to have rich-text rendering of comments and descriptions in JIRA. However, we have a large number of existing comments and descriptions. When we change the renderer of these fields, it completely mangles the existing content since they were not written with this renderer in mind.
It seems to me that there are two reasonable solutions:
1) Transform all the old content so that, when rendered with the wiki renderer, it appears the same as it did before. Lots of ways to do this sort of escaping.
2) Mark all the old content as plaintext and use the plaintext renderer with them, only applying the wiki renderer to new content.
Solution (2) seems safer, since it could be much more easily disabled if something goes wrong. With solution (1), if the migration has a bug, then we will have to roll back to a backup. More likely we won't notice the bug for quite a while and will need to retrieve just a few comments or JIRA descriptions from the backup, or edit them by hand.
Is there any official solution to this, or has anyone else tackled this migration? How did you approach it?