On Tue, Dec 6, 2011 at 2:57 PM, Harold Hjalmarson <[email protected]> wrote:
> Hello, > > I originally sent this message to Jim Schimpf after I found his paper on > Fossil. He suggested that I should post a question to this mailing list. > > I'm curious if Fossil preserves the modification (mtime) time of a file. > I'm aware that this feature is controversial in that code developers see no > reason for preserving this quantity. For example, the concern is that such > preservation interferes with make. > Exactly. Preserving mtime messes up makefile, which is very bad for developers. Fossil does keep track of when each file was checked in - not the mtime on disk, but the time when you typed "fossil commit" to add its changes to the repository. You can see those times in timelines of specific files such as this: http://www.fossil-scm.org/fossil/finfo?name=src/delta.c On the rare occasion when we want to know when a file was modified, we can consult those timelines. (Notice that we get to see all change times over the complete history of the file, not just the most recent change.) But Fossil does not have a mode whereby it sets the mtime of files on checkout. The information is there, and we could in threory add such a capability. But, because of the makefile problems you allude to, I am not particularly motivated to add such a feature. > > I want to preserve the modification time because it is meaningful to me. > If I import a file that I last changed in 2010, I become confused if it > shows up as being changed in December 2011. > > I notice that Jim Schimpf's paper mentions the preservation of the mtime > in a setting. However, it might mean that it is first changed during a > commit but then preserved after that first change. > I think this means that Fossil can use mtime to keep track of which files in a repository have changed since the last commit. This is faster than actually examining the content of all the files to find changes. > > I also searched the archives before sending this email to the list. I > found related items but I didn't see an answer to my specific question. > From what I did see, I may need to infer that the timestamp is the commit > time. This approach seems to destroy the mtime information. > > I note that the Wikipedia article about version control software > comparison says the timestamp preservation is "unknown" for Fossil. It also > says it is preserved for CVS but I think it's not preserved. > > At this time I'm preserving this information by putting my files in a tar > file. The modification times of the tar file are not saved but the > extracted directory and its files all have the correct times. > > By the way, I'm a potential user, a theoretical physicist with poor > computer science skills. > > Harry > > _______________________________________________ > fossil-users mailing list > [email protected] > http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users > > -- D. Richard Hipp [email protected]
_______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
