On Wed, Oct 5, 2011 at 12:19 AM, Eric <e...@deptj.eu> wrote: > > On Wed, October 5, 2011 6:14 am, Matt Welland wrote: > ... > > Every time pragmatism loses to philosophy someone, somewhere, is gonna > get > > screwed. > > But "pragmatic" decisions made by someone who doesn't understand the > philosophy/principles/rules can also cause problems, >
Trying to defend against ignorance by shackling the hands of the non-ignorant will likely alienate some the non-ignorant. Irrelevant, illegal, embarrassing or burdensome stuff will occasionally make it into a fossil repo and being able to remove such items is a hard requirement for me and I suspect for quite a few others. Fossil already has a reasonable compromise in this regard. > It is noble to have a philosophy of "don't rewrite history" but only to an > extent. Some obvious and perhaps not so obvious examples have been > mentioned > in this thread. "Those who rewrite history are condemned to repeat it" perhaps. > I see only the weakest of connections between this statement and reality in this context. > I think fossil has a nice balance here. It is possible to remove stuff but > it takes a little effort. Never deleting stuff is just silly. An record of > the past that stores irrelevant data is quite likely less useful than a > record that has been cautiously cleaned up. Define "irrelevant", including how you can be absolutely sure that it > applies to any particular case. > In any given domain the non-ignorant will know with very little effort the difference between relevant and irrelevant. Accidentally checking in a couple hundred megs of data from the wrong project into a fossil repo is almost guaranteed to happen someday by someone on our dev. team. That data is easily discerned as irrelevant to anyone on the project. People make mistakes. As the advocate and admin for fossil I need to be able to recover gracefully from those mistakes. I don't need the process of removing the data to be made easy but I do need it to be possible. > I'd personally like to see a mechanism that does the following: > > 1. Stops the object(s) from being propagated or received. > 2. Hides the objects(s) from view, (but can be enabled to show again) > 3. After a period of time, I think about a year, the data is removed > entirely. Not altogether a bad idea but, as you might expect, I would worry about how > to select the time interval, and I think the system should insist on > keeping a comment about the deletion. > Making the time interval configurable would be fine. A comment is going to be just noise and noise distracts but that is fine also. I did forget "# 4. Make sure the objects are not inadvertently referenced anywhere in fossil once hidden." Eric > > -- > ms fnd in a lbry > > _______________________________________________ > fossil-users mailing list > fossil-users@lists.fossil-scm.org > http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users >
_______________________________________________ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
