I haven't progressed those ideas yet with the main list. I'm just in the
process of updating my kit, plus a few familiy issue, which meant I've been
short on time.

I wanted to do some research on whether commit-tree would take the 00000
hash as being a valid parent (as a stand in for root) or whether that would
be a 'big ask'. Finding amethod for creating a sentinel value felt to be a
solution.

I suspect the list will just say that a second branch should be created
(similar to Git's maint branch) which collates those old tips, which would
be much easier to implement, but would make blame harder. It probably needs
more diagrams ;-)

----- Original Message ----- From: "Michael_google gmail_Gersten" <keybou...@gmail.com>
To: "Philip Oakley" <philipoak...@iee.org>
Cc: "Git Users" <git-users@googlegroups.com>
Sent: Friday, June 15, 2018 8:48 PM
Subject: Re: [git-users] Marking a commit parent as not displayed by default


So what sort of discussion came out of this on the main git development
list?

On Fri, May 25, 2018 at 3:23 PM, Philip Oakley <philipoak...@iee.org>
wrote:
Interesting blog, but I don't think it's really that workable with
respect
to having every commit within the rebase being merge linked back to the
older version.

However I do think that support is needed for marking a previous release
commit (probably also version tagged) as being a terminus point, and then
merging in that terminus commit at the start of the new rebased series.
(It's the problem GfW has)

repeating the layout in Michael's blog (hoping my mail agent doesn't ruin
the ascii art)

M0----M1----M2           M0----M1----M2
. \                        \           \
.  \               -->      \           B1'---B2'
.   \                        \
.    B1----B2                 B1----B2
Standard rebase


M0----M1----M2           M0----M1----M2--
. \                        \              \
.  \               -->      \         000--3way --B1'---B2'
.   \                        \            /
.    B1----B2                 B1---B2---T0

Rebase with terminus T0, and 3way merge
T0 contains the empty tree (everything gets off the bus at the terminus)
000, the 2nd parent, is a null root sentinel - makes this 3-way merge
quite
special.

With this layout it now becomes relatively easy to spot the sentinel base
marker of the endlessly rebased sequence B1-B2 as the master release
moves
forward, and the old series isn't even on the second parent. The use of
the
root commit as a spare sentinel may be contentious though.

Philip

----- Original Message ----- From: "Philip Oakley" <philipoak...@iee.org>
To: "Git Users" <git-users@googlegroups.com>
Sent: Friday, May 25, 2018 8:44 PM
Subject: Re: [git-users] Marking a commit parent as not displayed by
default



Michael,

I don't believe so, but it is probably worth re-raising on the Git
mailing
list. (Michael Haggerty is currently active as well)

I say that because @dscho is currently doing a better rebase that can
handle
merges and amend the flow logic within such merged series that are being
rebased.

https://public-inbox.org/git/cover.1524659287.git.johannes.schinde...@gmx.de/

He is doing this especially to supprt Git for Windows which also need
the
'historical' segments retained after every rebased of the GfW fixes that
go
on top of the upstream git. Having a method of marking such older parts
as
'uninteresting' would be great because these patches repeat themselves
over
and over again as on passes the 2nd parent point of the old relase would
be
great.

Now to read that old post in detail....

Philip

----- Original Message ----- From: "Michael" <keybou...@gmail.com>
To: <git-users@googlegroups.com>
Sent: Friday, May 25, 2018 8:22 PM
Subject: [git-users] Marking a commit parent as not displayed by default


A few years ago (2009), the idea was put forward that some parents in a
merge should be marked as "historical", or "uninteresting", and not
displayed by default

(http://softwareswirl.blogspot.com/2009/08/rebase-with-history-implementation.html)

Does git currently have any support for this?

---
Entertaining minecraft videos
http://YouTube.com/keybounce

--
You received this message because you are subscribed to the Google
Groups
"Git for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send
an
email to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.


--
You received this message because you are subscribed to the Google
Groups
"Git for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send
an
email to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.





--
Political and economic blog of a strict constitutionalist
http://StrictConstitution.BlogSpot.com

This message may have been spell checked by a laptop kitten.

--
You received this message because you are subscribed to the Google Groups
"Git for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.


--
You received this message because you are subscribed to the Google Groups "Git for 
human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to