Hi Soren,

On Sat, 22 Mar 2025 at 20:35, Soren Stoutner <so...@debian.org> wrote:
> On Saturday, March 22, 2025 1:09:29 PM Mountain Standard Time
> Christopher Obbard wrote:
> > Hi!
> >
> > For the package fluster, upstream at
> > http://github.com/fluendo/fluster/, we have a slight mistake with
> > the upstream versioning and I would like to add an epoch
> >
> > The current version in debian is 1.0.1+git20231211.d9106f5-1 (since
> > upstream had released 1.0.1 but later deleted the tag) and now
> > upstream latest version is 0.2.0+git20250319.30ac3a8
> >
> > I am proposing to add an epoch; e.g 1:0.2.0+git20250319.30ac3a8 to
> > make the mistakes from the past go away.
> >
> > Does it sound OK to others ?
>
> Does upstream plan to release a 1.0.1 sometime in the future?

I guess eventually upstream will release 1.0.1 but I don't think it
will be for a long while.
They had a few issues with their github release pipeline last year,
but all seems to be OK now.


> If so, I would probably go with a 1.0.1+really0.2.0 release until
> upstream catches up.
>
> Typically, I don’t like to go with an epoch unless upstream does
> something that will never be compatible, like switching from a date
> release of 2025-01-27 to 1.0.1.

I am happy to do that rather than add an epoch; does anyone else have
any opinions ?

I thought an epoch would be cleaner here personally, but I don't have
any actual upstream Debian
packaging experience with epochs. When developing custom distros with
custom (not suitable for Debian)
packages we added them fairly frequently when bootstrapping things and
upstream changing things.
But it seems like adding epochs in Debian is a much more conservative
decision? It'd be interesting to
hear more about this, I possibly need some history lesson for
real-world case where adding epoch
is dangerous ;-).


Cheers!

Chris

Reply via email to