Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.devel.release > #130652 > unrolled thread

Is a self-contained upgrade considered large/disruptive? (was Re: Bug#1102599: rust-tonic: please upgrade to 0.13 with upcoming upgrade of rust-prost 0.13)

Back to article view | Back to linux.debian.devel.release

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Is a self-contained upgrade considered large/disruptive? (was Re:  Bug#1102599: rust-tonic: please upgrade to 0.13 with upcoming upgrade of  rust-prost 0.13) Blair Noctis <ncts@debian.org> - 2025-04-11 17:20 +0200
    Re: Is a self-contained upgrade considered large/disruptive? (was Re:  Bug#1102599: rust-tonic: please upgrade to 0.13 with upcoming upgrade of  rust-prost 0.13) Paul Gevers <elbrus@debian.org> - 2025-04-12 12:50 +0200
      Re: Is a self-contained upgrade considered large/disruptive? (was Re:  Bug#1102599: rust-tonic: please upgrade to 0.13 with upcoming upgrade of  rust-prost 0.13) Blair Noctis <ncts@debian.org> - 2025-04-12 17:40 +0200

#130652 — Is a self-contained upgrade considered large/disruptive? (was Re: Bug#1102599: rust-tonic: please upgrade to 0.13 with upcoming upgrade of rust-prost 0.13)

[Multipart message — attachments visible in raw view] — view raw

Dear RT,

(Please consider this specific issue low priority,
but I think it's useful to clarify a general question.)

On 11/04/2025 03:44, Jonas Smedegaard wrote:
> Quoting Blair Noctis (2025-04-10 21:26:26)
>> I'm about to upgrade the prost family from 0.12 to 0.13.
> 
> That sounds like a "large or disruptive changes to unstable" which the
> release team as explicitly stated is not ok during freeze, so please
> refrain from doing that, or if needed then please ask the release team
> for an exception.

The affected packages are:

prost family, target of this upgrade:

- rust-prost-build
- rust-prost-types
- rust-prost
- rust-prost-derive

Dependents:

- rust-parsec-service
- rust-parsec-interface
- rust-pprof
- rust-tonic
- rust-axum
- rust-cast-sender
- rust-prometheus-client
- netavark
- gnome-authenticator

I have prepared all of them for this upgrade.
All build and pass tests.

I regret to have not noticed this chance and prepared earlier.
But:
I argue this is not "disruptive":
Updates of the prost family only affect listed dependents.
Updates of said dependents don't introduce breaking changes.
The whole upgrade is self-contained and does not "leak" to other packages.
Also not "large":
All considered, the "blast radius" is 13 packages.
This is fairly small IMO.
All rust-* packages besides rust-tonic and rust-axum are maintained in the Rust team,
of which I'm a member, thus in the position to handle.
In rust-tonic and rust-axum,
maintained by Jonas whose words are quoted above,
the former received a prepared upgrade for the former (#1102599, this bug),
and the latter requires only a version bump (#1102597).
netavark is maintained by Reinhard Tartler (siretart),
and gnome-authenticator by Jeremy Bícha (jbicha),
both also requiring only a version bump,
and both gave approval (netavark #1102598, g-a on IRC).
This is a fairly small circle.

So:
If all packages affected are handled ("self-contained"),
the blast radius is rather small,
and I have the go from all affected maintainers,
is it still considered large/disruptive?

Thanks.

Btw, when asked on IRC,
Ramacher said it depends on the size of debdiffs,
so here they are:
https://people.debian.org/~ncts/rust-prost-0.13-affected-debdiffs/
(Cargo.lock stripped as it's basically dead weight)

-- 
‌    ,Sdrager
Blair Noctis

[toc] | [next] | [standalone]

#130674

[Multipart message — attachments visible in raw view] — view raw

Hi Blair,

On 11-04-2025 17:14, Blair Noctis wrote:
> So:
> If all packages affected are handled ("self-contained"),


This is indeed good.

> the blast radius is rather small,


In transition context, we generally consider "small" when numbers are in 
the low single digit. Otherwise we'd want to be involved/informed. That 
always true, but more so during the transition freeze of course. By the 
way, our definition of transition is broader than the classical 
C-library transition, and covers changes to packages that require 
changes to or rebuilds of other packages.

> and I have the go from all affected maintainers,


That's obviously also good.

> is it still considered large/disruptive?


It's at least against the spirit of the freeze, see our definition of 
transition above.

> Ramacher said it depends on the size of debdiffs,
> so here they are:
> https://people.debian.org/~ncts/rust-prost-0.13-affected-debdiffs/
> (Cargo.lock stripped as it's basically dead weight)


If you want to request an exception, please file a bug and attach 
debdiffs to the report. That way they are stored in the BTS. Having said 
that, the size of those debdiffs don't look good. That's not really 
reviewable. So you better have a good explanation of why the new version 
is so much better than the current version in trixie. Otherwise consider 
this request as just too late.

Paul

[toc] | [prev] | [next] | [standalone]

#130687

[Multipart message — attachments visible in raw view] — view raw

Hi Paul,

On 12/04/2025 18:42, Paul Gevers wrote:
> Hi Blair,
> 
> On 11-04-2025 17:14, Blair Noctis wrote:
>> So:
>> If all packages affected are handled ("self-contained"),
> 
> 
> This is indeed good.
> 
>> the blast radius is rather small,
> 
> 
> In transition context, we generally consider "small" when numbers are in the low 
> single digit. Otherwise we'd want to be involved/informed. That always true, but 
> more so during the transition freeze of course. By the way, our definition of 
> transition is broader than the classical C-library transition, and covers 
> changes to packages that require changes to or rebuilds of other packages.

Thanks for the explanation and clarification.
Especially the "low single digit" one.
This is the main point of confusion here, I think.

>> and I have the go from all affected maintainers,
> 
> 
> That's obviously also good.
> 
>> is it still considered large/disruptive?
> 
> 
> It's at least against the spirit of the freeze, see our definition of transition 
> above.
> 
>> Ramacher said it depends on the size of debdiffs,
>> so here they are:
>> https://people.debian.org/~ncts/rust-prost-0.13-affected-debdiffs/
>> (Cargo.lock stripped as it's basically dead weight)
> 
> 
> If you want to request an exception, please file a bug and attach debdiffs to 
> the report. That way they are stored in the BTS. Having said that, the size of 
> those debdiffs don't look good. That's not really reviewable. So you better have 
> a good explanation of why the new version is so much better than the current 
> version in trixie. Otherwise consider this request as just too late.
I can't give a strong enough argument of "why the new version is so much better",
other than "it fixes some problems (*shows upstream changelog*)",
so it's probably not that important,
thus not worth an exception.

Thanks for the review!

-- 
‌    ,Sdrager
Blair Noctis

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.devel.release

csiph-web