We need to identify test cases and thoroughly test p2fs. We should also identify any and all architectural changes we want to make. Also, please take a look at 29182 to decide if it should be merged.
Also please decide whether or not merging 29180 is appropriate.
And 29056.
David, can you comment on the three changelists above (and any others that may be relevant?). Thanks.
@29182: Having PFS responsible for deleting empty live bundles and folders is really ugly. I had to put this back into WCMS since there is some pretty screwed up, DP specific code that depends on it (the DP-specific move functionality). For core CMS, it would be much cleaner if live bundles and folders were deleted when the last instance/item in them is unpublished. PFS doesn't really care either way. @29180: This invariant should be added to core CMS; if live and draft bundles/folders are not in the right position, horrible things will happen; in particular, publishing new items in them will make them appear in the wrong place on the live site. @29056: AFAIK, SystemWideReader wasn't pulled into core; it is very DP specific. It is used to log in as a special user which is then used to make the requests back to the editorial server from PFS. Core CMS might need something there if read access to live items on the editorial server is ever restricted via the public viewers group, but the code in this changelist is very DP-specific and most likely not useful for core CMS.
Invariant is merged (and ported to ANSI standard SQL so that it runs in both PG and Oracle).
Regarding the first paragraph of David's comment, change 31978 makes item unpublish recursively unpublish parents that are empty folders or empty bundles.