Re: [PATCH 4/4] promisor-remote: check advertised name or URL

[Junio C Hamano <gitster@pobox.com>]

Christian Couder <christian.couder@gmail.com> writes:

> A previous commit introduced a "promisor.acceptFromServer" configuration
> variable with only "None" or "All" as valid values.
>
> Let's introduce "KnownName" and "KnownUrl" as valid values for this
> configuration option to give more choice to a client about which
> promisor remotes it might accept among those that the server advertised.

A malicous server can swich name and url correspondence.  The URLs
this repository uses to lazily fetch missing objects from are the
only thing that matters, and it does not matter what name the server
calls these URLs as, I am not sure what value, if any, KnownName has,
other than adding a potential security hole.

> In case of "KnownUrl", the client will accept promisor remotes which
> have both the same name and the same URL configured on the client as the
> name and URL advertised by the server.

This makes sense, especially if we had updates to documents I
suggested in my review of [3/4].  If the side effect of "accepting"
a suggested promisor remote were to only use it as a promisor remote
on this side, there is no reason to "accept" the same thing again,
but because the main effect at the protocol level of "accepting" is
to affect the behaviour of the server in such a way that it is now
allowed to omit objects that are requested but would be available
lazily from the promisor remotes in the response, we _do_ need to
be able to respond with the promisor remotes we are willing to and
have been using.

This iteration does not seem to have the true server side support to
slim its response by omitting objects that are available elsewhere,
but I agree that it is a good approach to get the protocol support
right.

Thanks.