[Xastir-dev] Git question

Tom Russo russo at bogodyn.org
Mon Apr 22 09:19:39 PDT 2019 

Ok, I see what you have done.

This is what I have (my "origin" remote is what you have as upstream:

* 34f7373b (HEAD -> master, origin/master, origin/HEAD) Add dependent package
* e7f47a88 Rename OSX build
* 1247152f Fix overlooked unused result warning
* 6e14f284 Silence used result warning
* da55007d Silence gcc warnings about unused fread return result
* cb509f0f Remove three "stringop-overflow" warnings
* 1545996b (topic-removewarnings3) Remove some strncat stringop-overflow warning

Looks like you played around with the graphics magick issue by 
tinkering on your clone's master branch, which you then pushed to your 
fork to make Travis build it.

It looks like you then did a "git pull --rebase" or somesuch after you 
saw that I had committed and pushed to upstream/master.  By rebasing your
two commits on top of mine, you changed their SHA-1 hashes --- and therefore
you made your fork and the upstream repo diverge.

So:  don't do that.  Leave your fork's master alone, always, so it never
is anything but a copy of upstream's.  Do all work on your fork on side
branches, always.  If you stick to that, this won't happen.

Now, how to fix it:
 
I have already gotten upstream fixed for graphicsmagick, so your additional
commits are now moot.  So if I were in your position I'd force my clone back
to match upstream by discarding those commits altogether and force pushing 
upstream's master to origin:

    git checkout master
    git reset --hard upstream/master
    git push --force origin master

This will discard your "work to add graphicsmagick properly" and "second try
to get graphicsmagick into the build" commits, which should be unnecessary.

If you hate to lose that work even though it's moot, and *really* want to save 
those two commits, you can force them onto a side branch:

    git checkout master
    git checkout -b topic-savemycommits
    git checkout master
    git reset --hard upstream/master   
    git push --force origin master

The two commits are now only on your "topic-savemycommits" branch, to which
you can do anything you like, such as pushing to your fork.

The hard reset of master literally just moves the "master" ref back to point
to the same SHA-1 as upstream/master, and then checks out that SHA-1 (discarding
any changes to tracked files in the process).  When you did the checkout -b
up there, it simply created a new ref that points to the head of what
used to be your rebased master, thereby preventing git from garbage collecting
the commits when you reset.

And in the future, if you want to avoid problems like this, don't work on
master in your fork, ever.  Work on side branches.  This issue almost certainly
came from you making commits and pushing them to master on your branch,
then seeing my commit and doing a pull and doing some sort of thing (pull 
--rebase?) to get my commit into your branch *and then pushing the
resulting mangled master back to your fork*.

When I was working on getting graphicsmagick right, I was doing it on a
branch in my fork.  Only when I was actually sure it was working did I cherry
pick the one commit that did the trick back to master, and then pushed that
to Xastir/Xastir.  I never touched the master branch of my fork.

For the record, what I did to get graphicsmagick into the 16.04 build was:
   - make a new branch
   - on the branch, insert "cat config.log" after "configure" in the "script:"
     section
   - read config.log to see why it said it couldn't find WriteImage in 
     graphicsmagick libraries.  Find that it was bitching that it couldn't
     find "-lwebp".  
   - Discover that Ubuntu 16.04's graphicsmagick1-dev package is
     missing a dependency on libwebp-dev.  
   - Add that package to the list of packages to install on my branch
   - Confirm that it now works 
   - cherry-pick the "add that package" commit back to master and push to
     upstream.


On Mon, Apr 22, 2019 at 08:25:49AM -0700, we recorded a bogon-computron collision of the <curt.we7u at gmail.com> flavor, containing:
> I'll put my "gitv" output here to help Tom:
> 
> * 9d631f5f (HEAD -> master) Second try to get GraphicsMagick into the build
> * b0e2e82b Work to add GraphicsMagick properly
> * 34f7373b (upstream/master) Add dependent package
> | * 4f06ae45 (origin/master, origin/HEAD) Add dependent package
> | * b55a086c Second try to get GraphicsMagick into the build
> | * e6953131 Work to add GraphicsMagick properly
> |/
> * e7f47a88 Rename OSX build
> 
> 
> On Mon, Apr 22, 2019 at 8:24 AM Curt Mills <curt.we7u at gmail.com> wrote:
> 
> > So... I have my we7u/Xastir repo in this state:
> >
> > $ git pull upstream master
> > From github.com:Xastir/Xastir
> >  * branch              master     -> FETCH_HEAD
> > Current branch master is up to date.
> >
> > $ git status
> > On branch master
> > Your branch and 'origin/master' have diverged,
> > and have 3 and 3 different commits each, respectively.
> >   (use "git pull" to merge the remote branch into yours)
> > Untracked files:
> >   (use "git add <file>..." to include in what will be committed)
> > ...
> > nothing added to commit but untracked files present (use "git add" to
> > track)
> >
> > How do I easily resolve this? Do a "git pull origin master" and then see
> > what I end up with?
> >
> > What's the easiest way to see what the top three commits are on
> > origin/master?
> >
> > Using gitk I see this at the top:
> >
> > * master     Second try to get GraphicsMagick into the build
> > * Work to add GraphicsMagick properly
> > * remotes/upstream/master     Add dependent package
> > * Rename OSX build
> > * Fix overlooked unused result warning
> >
> > I don't need the top two commits: Tom Russo fixed that properly on
> > "upstream". They were attempts to fix something for Travis-CI and I had to
> > push them up to my origin/master to get Travis-CI to kick off to test them.
> >
> > --
> > Curt, WE7U        http://we7u.wetnet.net
> > http://www.sarguydigital.com
> >
> 
> 
> -- 
> Curt, WE7U        http://we7u.wetnet.net        http://www.sarguydigital.com
> _______________________________________________
> Xastir-dev mailing list
> Xastir-dev at lists.xastir.org
> http://xastir.org/mailman/listinfo/xastir-dev

-- 
Tom Russo    KM5VY
Tijeras, NM  

 echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]

More information about the Xastir-dev mailing list