At 07:52 AM 1/14/2004, The IETF Secretariat wrote:When an Internet-Draft expires, a "tombstone" file will be created that includes the filename and version number of the Internet-Draft that has expired. The filename of the tombstone file will be the same as that of the expired Internet-Draft with the version number increased by one. If a revised version of an expired Internet-Draft is submitted for posting, then the revised version will replace the tombstone file and will receive the same version number as that previously assigned to the tombstone file. Tombstone files will never expire and will always be available for reference unless they are replaced by updated versions of the subject Internet-Drafts.
Does this also apply in the "published as RFC" and "replaced by draft-foo" cases, not just expiration without publication?
If I can have two separate files (a tombstone and a subsequent new file version) that have the same name, as described in the recent announcement, I am going to have to figure out a trigger that will tell me that I need to re-download the file.
Kind of like how 1id-*.txt changes almost daily? If you can't use the timestamp as a trigger, the file size would probably be fairly reliable. For my personal mirror, I use rsync, and I've noticed no problems.
From what I've been told, the handling of numbers described above has been the procedure for a while. I asked about it a year or so ago and was told that the number of the tombstone file should be reused for the next submission "reviving" the draft. So, you may have been missing updates for a while.
It seems to me that there is a better approach to the above, at least in the context of the above. If the "tombstone" is literally as described, it would be far more space/search/etc efficient for us to have the tombstone consist of an added text line in a file indicating that the named draft expired on a certain date, and keep separate files for the active internet drafts. It seems to me that this makes it simpler to maintain a mirror and to find temporary documents.
Keeping the tombstone files around forever does sound messy. I like your suggestion of using a file to collect the data, at least for the long term. We'd want the author contact (and disposition, if not just for expiration) info around for at least a while, though maybe not permanently. So maybe still keeping tombstone files for six months or so would be wise. And it's the established way of finding out what's happening with recent enough drafts.
This file with the "used" names might grow to be pretty large after a lot of years, but we're probably not talking about more than a megabyte or two a year, are we, even if we keep all the author info around? By the time it's big enough to be a problem, we'll probably have better solutions....
Ken