Re: [PATCH 11/12] receive-pack: treat namespace .have lines like alternates

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Jan 27, 2017 at 06:45:26PM +0100, Lukas Fleischer wrote:

> > This is an unrelated tangent, but there may want to be a knob to
> > make the code return here without even showing, to make the
> > advertisement even smaller and also to stop miniscule information
> > leakage?  If the namespaced multiple projects are totally unrelated
> > (i.e. "My sysadmin gave me a write access only to this single
> > directory, so I am using the namespace feature to host these three
> > projects that have nothing to do with each other"), showing objects
> > of other namespaces will buy us nothing and the user is better off
> > without this code showing these refs as ".have".
> 
> I think this is already possible using receive.hideRefs (which causes
> the ref_is_hidden() branch above to return if applicable).
> 
> Having support for suppressing .have lines corresponding to different
> namespaces was actually the reason I implemented 78a766ab6 (hideRefs:
> add support for matching full refs, 2015-11-03). We have been using
> namespaces for hosting the package Git repositories of the Arch Linux
> User Repository [1] with a shared object storage for several months now.
> See [2] for *some* technical details on how things are implemented; the
> last section explains how the hideRefs mechanism can be used to limit
> ref advertisement to the "active" namespace.

Thanks for the pointers. I think a "turn off namespace .have lines"
option would be easier for some cases, but what you've implemented is
much more flexible. So if people using namespaces are happy with it, I
don't see any need to add another way to do the same thing.

-Peff



[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]