© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
73416ba
(Get the code!)
> It doesn't cause loss of history but it does cause us to store files as changed when they're not really, which makes
> things slow, and makes the graph history too long. So it is pretty bad.
I would like to say that it is more than bad: it effectively causes the destruction (at high speed, in our case in a month) of a repository (in my view pulls/pushes of over 2 hours, and commits taking half an hour mean your repository is unusable), and because it cannot be undone this means the only way out is to recreate everything from scratch, loosing all history in the process. In my view that is disastrous. The impact of this problem on a team of about 15 man has been quite big in terms of work hours lost. I certainly hope no one else currently goes live with crlf content filtering....
I still need to test the fix (will do it, but later); this discussion and some earlier indications I got from jam made me very unsure whether this fix is something to try on production; I currently put a lot of effort of testing the 2a format instead.