2.43 release date adjustment?

2.43 release date adjustment?

[Taylor Blau]

Now that we're on the other side of 2.42, it looks like the tentative
release date for 2.43 has been scheduled for 2023-11-20.

That week is a US holiday (Thanksgiving, on 2023-11-23). I figure that
many US-based folks may take a good portion of if not the whole week
off.

Should we consider pushing back the release to the week after
Thanksgiving, say on 2023-11-27? This cycle is already somewhat long so
we could move the release up to 2023-11-13, but I think it would be
nice to have folks around in case of any brown paper-bag fixes.

Thanks,
Taylor

Re: 2.43 release date adjustment?

[Junio C Hamano]

Taylor Blau <me@ttaylorr.com> writes:

> Now that we're on the other side of 2.42, it looks like the tentative
> release date for 2.43 has been scheduled for 2023-11-20.
>
> That week is a US holiday (Thanksgiving, on 2023-11-23). I figure that
> many US-based folks may take a good portion of if not the whole week
> off.

True.

> Should we consider pushing back the release to the week after
> Thanksgiving, say on 2023-11-27? This cycle is already somewhat long so
> we could move the release up to 2023-11-13, but I think it would be
> nice to have folks around in case of any brown paper-bag fixes.

I am not sure how it would help to place the release day immediately
after the off-week everybody has stayed off the keyboard, though.  I
would prefer to see fully working weeks during the pre-release
period so that people can work on last minute regression fixes, so I
hoped it would work much better to release before the holiday week
begins, rather than after it ends.

But of course I am open to inputs.  FYI, I plan to be offline _after_
the Thanksgiving week.

Re: 2.43 release date adjustment?

[Taylor Blau]

On Wed, Aug 23, 2023 at 09:20:33AM -0700, Junio C Hamano wrote:
> > Should we consider pushing back the release to the week after
> > Thanksgiving, say on 2023-11-27? This cycle is already somewhat long so
> > we could move the release up to 2023-11-13, but I think it would be
> > nice to have folks around in case of any brown paper-bag fixes.
>
> I am not sure how it would help to place the release day immediately
> after the off-week everybody has stayed off the keyboard, though.  I
> would prefer to see fully working weeks during the pre-release
> period so that people can work on last minute regression fixes, so I
> hoped it would work much better to release before the holiday week
> begins, rather than after it ends.

Perhaps then we should shorten this cycle by a week and plan to cut the
release on 2023-11-13.

That gives us a full working week to catch any bugs, and gives everybody
a nice break before starting on 2.44. Thoughts?

Thanks,
Taylor

Re: 2.43 release date adjustment?

[Junio C Hamano]

Taylor Blau <me@ttaylorr.com> writes:

> Perhaps then we should shorten this cycle by a week and plan to cut the
> release on 2023-11-13.

I still do not see the point.  Do you mean you do not want to work
between 13th to 17th of November that would be the "final regression
hunt in -rc2" week?

If you have a specific reason to expect a post-release bug hunt,
that would change the equation, but otherwise, I am not sure what
you are trying to achieve, compared to the original plan to tag on
20th, give developers week off after the release for Thanksgiving or
whatever the excuse to take a week off in your parts of the world.

The users will be taking the holiday week and will be slow after the
release anyway, and developers work should also happen more before
the release and not after it, no?

Re: 2.43 release date adjustment?

[Taylor Blau]

On Wed, Aug 23, 2023 at 02:39:41PM -0700, Junio C Hamano wrote:
> Taylor Blau <me@ttaylorr.com> writes:
>
> > Perhaps then we should shorten this cycle by a week and plan to cut the
> > release on 2023-11-13.
>
> I still do not see the point.  Do you mean you do not want to work
> between 13th to 17th of November that would be the "final regression
> hunt in -rc2" week?

That wasn't quite what I was saying, but maybe I am just confused by
what you mean.

I am suggesting cutting the release on the 13th of November instead of
the 20th so that we have folks online for the first week after the
release tag is pushed out to fix any brown-paper-bag issues.

Then, after the release is no longer fresh, we can expect many US
developers to take the week of the 20th off.

> If you have a specific reason to expect a post-release bug hunt,
> that would change the equation, but otherwise, I am not sure what
> you are trying to achieve, compared to the original plan to tag on
> 20th, give developers week off after the release for Thanksgiving or
> whatever the excuse to take a week off in your parts of the world.

No specific reason to expect bugs. Just trying to line things up so that
we have (a) people around when the release is tagged, and (b) don't have
a release hanging over our heads either during or immediately after a
week when most US developers are offline.

Thanks,
Taylor

Re: 2.43 release date adjustment?

[Junio C Hamano]

Taylor Blau <me@ttaylorr.com> writes:

> No specific reason to expect bugs. Just trying to line things up so that
> we have (a) people around when the release is tagged, and (b) don't have
> a release hanging over our heads either during or immediately after a
> week when most US developers are offline.

That is exactly what I call "expecting a post-release fixes".

A much more important thing is to make sure we catch them before the
release and devise ways to help us do so, perhaps opting more userbase
into testing early.

Also a much more productive thing to discuss than just the timeline:
What do we want the end product of this cycle to include?

I am personally hoping to see an enhanced rerere that can record and
replay delete-modify conflict that gets resolved to removal.  Others
may have their own agenda, and it may be interesting to hear them.
Those without specific agenda for this cycle may volunteer to help
others' topics.

Thanks.

Re: 2.43 release date adjustment?

[Taylor Blau]

On Wed, Aug 23, 2023 at 03:03:45PM -0700, Junio C Hamano wrote:
> Taylor Blau <me@ttaylorr.com> writes:
>
> > No specific reason to expect bugs. Just trying to line things up so that
> > we have (a) people around when the release is tagged, and (b) don't have
> > a release hanging over our heads either during or immediately after a
> > week when most US developers are offline.
>
> That is exactly what I call "expecting a post-release fixes".
>
> A much more important thing is to make sure we catch them before the
> release and devise ways to help us do so, perhaps opting more userbase
> into testing early.

I agree, but shouldn't we still plan such that we have folks around for
a week after the release in case something was missed?

Thanks,
Taylor

Re: 2.43 release date adjustment?

[Junio C Hamano]

Taylor Blau <me@ttaylorr.com> writes:

> On Wed, Aug 23, 2023 at 03:03:45PM -0700, Junio C Hamano wrote:
>> Taylor Blau <me@ttaylorr.com> writes:
>>
>> > No specific reason to expect bugs. Just trying to line things up so that
>> > we have (a) people around when the release is tagged, and (b) don't have
>> > a release hanging over our heads either during or immediately after a
>> > week when most US developers are offline.
>>
>> That is exactly what I call "expecting a post-release fixes".
>>
>> A much more important thing is to make sure we catch them before the
>> release and devise ways to help us do so, perhaps opting more userbase
>> into testing early.
>
> I agree, but shouldn't we still plan such that we have folks around for
> a week after the release in case something was missed?

I would rather want to see us try hard to avoid regressing in the
first place, instead of plan to have regression in the release and
have folks around after the release date.