threads / patch / 66632

patchdoc: clarify promisor remote fetch ordering

Subject: [Outreachy PATCH] doc: clarify promisor remote fetch ordering

## tl;dr

4 messages between Oct 22, 2026 and Oct 26, 2026. Diffs are folded; open one to read it.

replies: 3people: 2as markdown or json

Wasswa Edson· Oct 22, 2026, 11:02 UTC · lore
From fe5f35b00eafa1eb2756178445d3e88c4e9d6787 Mon Sep 17 00:00:00 2001
From: Edson <edsonwasswa54@gmail.com>
Date: Thu, 22 Oct 2026 13:09:21 +0300
Subject: [PATCH] doc: clarify promisor remote fetch ordering
Signed-off-by: Edson <edsonwasswa54@gmail.com>
---
 Documentation/technical/partial-clone.adoc | 5 +++++
 1 file changed, 5 insertions(+)
Show changes to Documentation/technical/partial-clone.adoc +5 −1
diff --git a/Documentation/technical/partial-clone.adoc
b/Documentation/technical/partial-clone.adoc
index 18718a3840..a1104930fb 100644
--- a/Documentation/technical/partial-clone.adoc
+++ b/Documentation/technical/partial-clone.adoc
@@ -196,6 +196,11 @@ filtered `git-fetch` commands from the central server.

 When fetching objects, promisor remotes are tried one after the other
 until all the objects have been fetched.
+When a server advertises promisor remotes via the protocol v2
+"promisor-remote" capability and the client accepts them (see
+linkgit:gitprotocol-v2[5]), those accepted remotes are tried first.
+After trying the accepted remotes, Git then tries the other configured
+promisor remotes as described below.

 Remotes that are considered "promisor" remotes are those specified by
 the following configuration variables:
--
2.53.0
Junio C Hamano· Oct 22, 2026, 17:35 UTC · re: Wasswa Edson · lore

Re: [Outreachy PATCH] doc: clarify promisor remote fetch ordering

Wasswa Edson <edsonwasswa54@gmail.com> writes:
> From fe5f35b00eafa1eb2756178445d3e88c4e9d6787 Mon Sep 17 00:00:00 2001
> From: Edson <edsonwasswa54@gmail.com>
> Date: Thu, 22 Oct 2026 13:09:21 +0300
> Subject: [PATCH] doc: clarify promisor remote fetch ordering

None of the above should be included in the e-mailed patch. Your MUA would record your name in the "From:" header, the commit title in the "Subject:" header, and the date when the public first saw this contribution from you in the "Date:" header, which should become the author date recorded in the project history.

On the other hand, there needs to be an explanation to convince the project that it should want this change before your sign-off.

> Signed-off-by: Edson <edsonwasswa54@gmail.com>
See Documentation/SubmittingPatches [[dco]] and [[real-name]].
Show 14 quoted lines
> diff --git a/Documentation/technical/partial-clone.adoc
> b/Documentation/technical/partial-clone.adoc
> index 18718a3840..a1104930fb 100644
> --- a/Documentation/technical/partial-clone.adoc
> +++ b/Documentation/technical/partial-clone.adoc
> @@ -196,6 +196,11 @@ filtered `git-fetch` commands from the central server.
>
>  When fetching objects, promisor remotes are tried one after the other
>  until all the objects have been fetched.
> +When a server advertises promisor remotes via the protocol v2
> +"promisor-remote" capability and the client accepts them (see
> +linkgit:gitprotocol-v2[5]), those accepted remotes are tried first.
> +After trying the accepted remotes, Git then tries the other configured
> +promisor remotes as described below.
Just fact checking.

I thought that a remote Git server cannot arbitrarily inject, advertise, or force a brand-new, untrusted promisor remote config onto a client during a protocol exchange (such as a standard git fetch or git clone) if the client doesn't already have it configured locally. But the above description makes it sound as if the server can advertise a remote repository that the client does not know or care about and direct the client to go there first. Is that true (yes, I am wondering if there is a security implication around here)?

Thanks.
Wasswa Edson· Oct 25, 2026, 23:35 UTC · re: Junio C Hamano · lore

Re: [Outreachy PATCH] doc: clarify promisor remote fetch ordering

Thanks so much about that but if i look into the codebase, Your concern is valid from a security perspective, but the implementation already has safeguards that prevent arbitrary injection Then about the Git server my say is a No, coz a server cannot arbitrarily inject a brand-new, untrusted promisor remote onto a client by default. The client must explicitly configure acceptance policies before any server-advertised remote will be accepted. If I am wrong let me be guided. I am re-doing the patch again following the doc you have provided Thank you

On Thu, 22 Oct 2026 at 20:35, Junio C Hamano <gitster@pobox.com> wrote:
Show 50 quoted lines
>
> Wasswa Edson <edsonwasswa54@gmail.com> writes:
>
> > From fe5f35b00eafa1eb2756178445d3e88c4e9d6787 Mon Sep 17 00:00:00 2001
> > From: Edson <edsonwasswa54@gmail.com>
> > Date: Thu, 22 Oct 2026 13:09:21 +0300
> > Subject: [PATCH] doc: clarify promisor remote fetch ordering
>
> None of the above should be included in the e-mailed patch.  Your
> MUA would record your name in the "From:" header, the commit title
> in the "Subject:" header, and the date when the public first saw
> this contribution from you in the "Date:" header, which should
> become the author date recorded in the project history.
>
> On the other hand, there needs to be an explanation to convince
> the project that it should want this change before your sign-off.
>
> > Signed-off-by: Edson <edsonwasswa54@gmail.com>
>
> See Documentation/SubmittingPatches [[dco]] and [[real-name]].
>
> > diff --git a/Documentation/technical/partial-clone.adoc
> > b/Documentation/technical/partial-clone.adoc
> > index 18718a3840..a1104930fb 100644
> > --- a/Documentation/technical/partial-clone.adoc
> > +++ b/Documentation/technical/partial-clone.adoc
> > @@ -196,6 +196,11 @@ filtered `git-fetch` commands from the central server.
> >
> >  When fetching objects, promisor remotes are tried one after the other
> >  until all the objects have been fetched.
> > +When a server advertises promisor remotes via the protocol v2
> > +"promisor-remote" capability and the client accepts them (see
> > +linkgit:gitprotocol-v2[5]), those accepted remotes are tried first.
> > +After trying the accepted remotes, Git then tries the other configured
> > +promisor remotes as described below.
>
> Just fact checking.
>
> I thought that a remote Git server cannot arbitrarily inject,
> advertise, or force a brand-new, untrusted promisor remote config
> onto a client during a protocol exchange (such as a standard git
> fetch or git clone) if the client doesn't already have it configured
> locally.  But the above description makes it sound as if the server
> can advertise a remote repository that the client does not know or
> care about and direct the client to go there first.  Is that true
> (yes, I am wondering if there is a security implication around
> here)?
>
> Thanks.
>
Junio C Hamano· Oct 26, 2026, 05:18 UTC · re: Wasswa Edson · lore

Re: [Outreachy PATCH] doc: clarify promisor remote fetch ordering

Wasswa Edson <edsonwasswa54@gmail.com> writes:
Show 6 quoted lines
> ... the implementation
> already has safeguards that prevent arbitrary injection
> Then about the Git server my say is a No, coz a server cannot
> arbitrarily inject a brand-new, untrusted promisor remote onto a
> client by default. The client must explicitly configure acceptance
> policies before any server-advertised remote will be accepted.

An explanation like the above must be in your proposed log message so that the next person who looks at your patch do not have to ask the same question.

Also, we frown upon top-posting on this list.
Thanks.

← back to recent threads