Forgot to mention: while watching the execution of the clients in a terminal, I saw that for some reason my ssh connection to the git server timed out and seemed to have difficulty reconnecting. But, I'm unclear why this connection should stay open in the first place.
The client seems to check at 10 minute intervals.
Reading up on Git, which I still do not fully understand, it sounds as though the SparkleShare choice to use Git makes good sense. If I'm not mistaken, the client's task is to automate making git command calls that a programmer might make when modifying code and commiting files. So, the client then needs to monitor whether any files have changed and commit the changes to the repository.
However, I am still interested in the windows-based synchronizer I mentioned previously for the reason that it can synchronize open files -- and this would be necessary if working with truecrypt containers. Don't know if the git synchronizer would do this.
But, another option would be to put both repositories in truecrypt containers, one at each end. These are engineering issues well beyond me -- but that is why the Git system sounds like a pretty good foundation: written by Linus Torvalds, it benefits from being based on a filesystem level programming. I'm thinking this is a wheel I should not try to reinvent.
Ok, so let's say I want to use Git as my file synchronizing mechanism. (Leaving aside the question of whether it will become size-prohibitive as apparently it keeps all prior versions ... but again, that has the benefit of not losing files when a screw-up occurs?)
If doing this manually without benefit of some slick front-end, what would I have to do?
- check the central server occasionally to see if another computer/user has made changes to any files and if so update the local git repository (pull) to match.
- whenever I change a file locally, commit that change to my local repository and synchronize (push) that to the central server.
So far, this is sounding pretty simple. Devil is in the details, I guess.
digression: Google just "gets it". My employer just started using Google Apps. The site builder is really easy, especially compared to this blogging software and joomla, etc. A nice, useable website doesn't need to be technology heavy. It just needs to be simple and accessible. Most software comes down to developers showing off their skills, often in cludgy programming languages that execute way too slowly. What really need is simplicity and usability.
Google comes much closer to the original Berners-lee conception of the web -- that you could readily add to the web, a more interactive concept than has developed. Of course, I am a luddite, in some ways I like the original text based web better than the modern one.
It's ironic to me that Google is both being very successful and achieving easy usability. Lately I have been thinking of going back to the style of webpages I built in 1995-6 for that reason. I don't think my "modern" webpages are more useable, just slower -- and won't be very useful on a smartphone.
So, I loaded a lot of "git" related software out of OpSu YAST. Still trying them out. gitg allows viewing of the git repository (eg., the sparkleshare repository). But qgit seems more useful in the same role, because it allows me to find and save files ... eg., makes it possible to retrieve an earlier version of the same file, or if I inadvertently deleted one, looks like I could get it back.
There are all sorts of utilities, such as web servers, git servers, etc. I don't have any idea what I need or what might be useful.
Trackback address for this post
No feedback yet
|« i branch to qt||sparkleshare is semi, sort of working »|