RFC: set timestamps on files created by bzr

Robert Collins robertc at robertcollins.net
Wed Jan 23 09:22:51 GMT 2008


On Wed, 2008-01-23 at 15:01 +1100, Martin Pool wrote:
> On 22/01/2008, Robert Collins <robertc at robertcollins.net> wrote:
> 
> > If bzr were to set a timestamp on any files it alters that is the same
> > for all files altered during the same operation, then timestamp-based
> > build systems that have these issues would not trigger rebuilds
> > incorrectly allowing wide scale deployment.
> 
> So they'd all be set, say, to the time we started doing the change?

Yes.

> It's an interesting point.  I can't think of any problems with doing
> it.  Would you just have it always turned on?

Yes.

> > This could be done by calling futime (which is not exposed in python
> > AFAICT, we'd need an extension to make it available) during writing the
> > files in tree transform.
> 
> Why isn't os.utime enough?

Its likely more efficient to do futime while the file is open; and this
is performance sensitive code - adding one syscall per path that will
require path->inode mapping in the kernel is pretty certain to have
measurable cost.

-Rob
-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20080123/f9611238/attachment.pgp 


More information about the bazaar mailing list