On Nov 10, 2007 12:42 PM, Xavier <shiningxc@gmail.com> wrote:
On Sat, Nov 10, 2007 at 07:14:11PM +0100, Nagy Gabor wrote:
Hi!
To be honest, I don't like the rollback solution; the pre-transaction checking is much more elegant (for me).
However, it's hard to tell how difficult it would be to go the rollback way before actually doing it. Probably a poor implementation of it would result in a worse situation than the current one.
This was my first instinct - we don't actually _have_ a "rollback solution" - we have talk of such a solution, and lots of speculation about how inelegant it is, with no actual implementation. I don't think either Dan or myself were thinking of getting rid of pre-checking for existing files. That's not complex. There's a difference between an expected error, and an unexpected one. Files existing - that's expected. The disk filling up WHILE installing (after the check is done), or hitting immutable files (ext FS only), these we can't account for. There will always be something that would cause an install to break, and codifying all cases is impossible. Recovering sanely, however, that is possible.