floatplane / git_pr Goto Github PK

View Code? Open in Web Editor NEW

6.0 6.0 3.0 72 KB

A command-line utility for creating, viewing, diffing, and merging GitHub pull requests

License: MIT License

Ruby 100.00%

git_pr's People

Contributors

Stargazers

Watchers

Forkers

petersibley vfleurima jasonreisman

git_pr's Issues

Restore original branch if possible.

It’d be nice if git_pr left you on the same branch where you were when you invoked it.

ie. i’m on feature/a and I perform git pr merge 123 (PR 123 corresponds to feature/a).

git pr fails, ie.:

Local branch 'feature/syncPageAttributes' differs from remote branch 'charlesmchen/feature/syncPageAttributes'. Please reconcile before continuing.

the temptation is to git push -f since you think you are still on feature/a.

I always have to catch myself.

Handle binary data being occasionally dumped out by `git diff`

bug report:

peter [5:01 PM] 
this is a new thing from `git-pr`  
/Library/Ruby/Gems/2.0.0/gems/git-1.2.8/lib/git/diff.rb:121:in `split': invalid byte sequence in UTF-8 (ArgumentError)
    from /Library/Ruby/Gems/2.0.0/gems/git-1.2.8/lib/git/diff.rb:121:in `process_full_diff'
    from /Library/Ruby/Gems/2.0.0/gems/git-1.2.8/lib/git/diff.rb:107:in `process_full'
    from /Library/Ruby/Gems/2.0.0/gems/git-1.2.8/lib/git/diff.rb:64:in `each'
    from /Library/Ruby/Gems/2.0.0/gems/git_pr-0.0.13/lib/git_pr/merge.rb:55:in `any?'
    from /Library/Ruby/Gems/2.0.0/gems/git_pr-0.0.13/lib/git_pr/merge.rb:55:in `merge_pull_cleanly'
    from /Library/Ruby/Gems/2.0.0/gems/git_pr-0.0.13/bin/git-pr:217:in `<top (required)>'
    from /usr/bin/git-pr:23:in `load'
    from /usr/bin/git-pr:23:in `<main>'

[5:01] 
when I run
~/gitclients/Paper (develop) $ git pr merge 4851

solution:

brian [8:58 AM] 
ok so your git_pr deal was this: your branch was pretty diverged from origin/develop (a manual rebase would have made this go away). one of the files that was different in the branches was `lib/FiftyThreeSDK/docs/FiftyThree-Trademark-Guidelines/FiftyThree-Trademark-Guidlines.pdf`, which has embedded binary parts. somehow git didn't recognize it as binary, so `git diff` was printing binary data instead of just saying "Binary file changed"

[8:59] 
now i know for next time.

peter [8:59 AM] 
Hah

[9:00] 
Odd - yeah that explains the gigantic binary thing that it spit out

Consider removing PR description from merge commit message

The JIRA integration didn't work, and it causes @-mentioned people to get emailed twice.

Add build status to list

diff and difftool should use the current directory if they can

like everything else does.

Restore checked out branch state when aborting

from @charlesmchen

Twice now I've had "git pr" fail with a message like this:

Local branch (feature/extractEntitlements) differs from remote branch (charlesmchen/feature/extractEntitlements). Please reconcile before continuing.

Since I started "git pr" while "feature/extractEntitlements" was the
current branch, I would respond (trying to resolve this issue) with:

git push -f

Of course, this was wrong - "git pr" had checked out "develop", so I
was now force-pushing develop. Dangerous! Both times, I was lucky
and had an up-to-date develop.

My request: "git pr" notes the current branch when it starts and if
there is an error, returns to that branch before exiting.

diff/difftool base not always correct

Problem seems to be that remote tracking branch for origin/develop is out of date.
Current branch is matthew/feature/syncActionCompositesRevisted. Head of origin/develop is 32d4484 but we are diffing against 17bc0f8 instead.

$ git lol
* e324fc1 (HEAD, matthew/feature/syncActionCompositesRevisted) Clean up before PR.
* dd95c3a Avoid crashes in PageStorage.
* 2d5bd58 Lazy-generate image composites for sync operations and snapshots.
*   32d4484 Merge #5304 from DawnWright: Search space titles
|\  
| * b91d2a7 Incorporate CR feedback from Bridger.
| * 9c42738 Update cell styling, remove subtitle.
| * 0c2e899 Slide top of header away & cancel in when searching
| * 926ece0 Refine search field behavior and add cancel.
| * ac248e7 Add search to the top-level navigation drawer.
|/  
*   17bc0f8 (origin/develop, origin/HEAD, develop) Merge #5303 from bridger: Fix layout crash from PhotoView race. PPR-5544 #resolve
$ git pr -v dt 5307
Opening standard repository in /Users/brian/src/Paper
GitHub project: FiftyThree/Paper
Getting pull request information...
Executing git difftool 17bc0f8ec890853846de76033b942dfdf9607fff matthew/feature/syncActionCompositesRevisted

Better detection of remote to query

Maybe look at both upstream and origin and figure out which is a fork of the other...?

[10:46 AM] Peter Sibley: random git pr question, git pr list appears to work in one of my workspaces but not in another.
[10:46 AM] Peter Sibley: Are credentials stored in my home dir or in the repo?
[10:47 AM] Brian Sharon: in your home dir, but if it's not prompting you for credentials then it's not that
[10:47 AM] Peter Sibley: I tried doing gem upadte stuff
        and still repros.
[10:47 AM] Peter Sibley: Is there a verbose mode or anything?
[10:47 AM] Brian Sharon: git pr -v list
[10:48 AM] Peter Sibley: ~/gitclients/Paper (develop) $ git pr -v list        No open pull requests found.
[10:48 AM] Brian Sharon: Also check the output of "git config --get pr.defaultremote" in both places
        it's probably the upstream vs origin thing
[10:48 AM] Peter Sibley: yup that's it

[10:50 AM] Peter Sibley: OK that fixed it. Thanks.
[10:50 AM] Brian Sharon: cool. i wish i had a better heuristic for automatically determining the "root" repo
[10:51 AM] Peter Sibley: is upstream configured if so use the origin + upstream pattern
        But agreed that it's confusing.
[10:52 AM] Brian Sharon: maybe i should check to see if upstream is a fork of origin or vice versa

open doesn't work on read-only origins

/Users/brian/src/json_schema [floatplane-fix-test] $ git pr o
Specified remote 'origin' is not a GitHub remote.
/Users/brian/src/json_schema [floatplane-fix-test] $ git remote -v
floatplane  [email protected]:floatplane/json_schema (fetch)
floatplane  [email protected]:floatplane/json_schema (push)
origin  https://github.com/brandur/json_schema (fetch)
origin  https://github.com/brandur/json_schema (push)

git pr open is very slow if no pull request exists

Not sure why, "-v" prints nothing useful.

Doesn't work from a submodule

[email protected]: /Users/brian/src/Paper/iOS/ThirdParty/InAppSettingsKit [feature/fix-static-analyzer-bugs] $ git pr o
/Users/brian/.gem/ruby/2.0.0/gems/git-1.2.8/lib/git/lib.rb:342:in `chdir': Not a directory - /Users/brian/src/Paper/iOS/ThirdParty/InAppSettingsKit/.git (Errno::ENOTDIR)
    from /Users/brian/.gem/ruby/2.0.0/gems/git-1.2.8/lib/git/lib.rb:342:in `config_get'
    from /Users/brian/.gem/ruby/2.0.0/gems/git-1.2.8/lib/git/base.rb:137:in `config'
    from /Library/Ruby/Gems/2.0.0/gems/git_pr-0.0.9.pre0/lib/git_pr/github.rb:85:in `determine_project_name_from_command_line'
    from /Library/Ruby/Gems/2.0.0/gems/git_pr-0.0.9.pre0/bin/git-pr:196:in `<top (required)>'
    from /usr/bin/git-pr:23:in `load'
    from /usr/bin/git-pr:23:in `<main>'

don't fetch twice if PR is against same project
difftool/diff
open

Verify that the merge looks clean:

*   5186863 (HEAD -> develop) Merge #5089 from floatplane: fix basic sync tests
|\  
| * de0a939 (floatplane/feature/fix-basic-sync-tests, feature/fix-basic-sync-tests-rebase, feature/fix-basic-sync-tests) CR feedback
| * 0a27856 Fix BasicSampleClientLibTests to cope with latest server changes
|/  
* a2da56e (origin/develop, origin/HEAD) Merge #5080 from mliberman: New Idea Button Refactor PPR-5156

Do you want to proceed with the merge (y/n)? y
Pushing changes to 'origin'

And then you're left with a merge commit that you have to unwind:

$ git lola
*   5186863 (develop) Merge #5089 from floatplane: fix basic sync tests
|\  
| * de0a939 (HEAD -> feature/fix-basic-sync-tests, floatplane/feature/fix-basic-sync-tests) CR feedback
| * 0a27856 Fix BasicSampleClientLibTests to cope with latest server changes
|/  
*   a2da56e (origin/develop, origin/HEAD) Merge #5080 from mliberman: New Idea Button Refactor PPR-5156

$ git co develop 
$ git reset --hard origin/develop

Catch case where branch is pushed with no tracking info

brian [12:51 PM] 
ok, let's try this:
$ git for-each-ref --format='%(upstream:short)' refs/heads/feature/shapeSampleRate3
mattfeizli/feature/shapeSampleRate3

mattfeiszli [12:53 PM] 
is that second line the output that you get, or is that part of the command?

brian [12:53 PM] 
output

mattfeiszli [12:53 PM] 
ok, i see this:

brian [12:53 PM] 
but yours should say "matt"

mattfeiszli [12:53 PM] 
~/gitclients/PaperSDKCore (feature/shapeSampleRate3) $ git for-each-ref --format='%(upstream:short)' refs/heads/feature/shapeSampleRate3

~/gitclients/PaperSDKCore (feature/shapeSampleRate3) $

brian [12:54 PM] 
i don't think your branch has tracking info set. try this: `git push -u origin feature/shapeSampleRate3`

[12:54] 
("-u" is shorthand for "--set-upstream-to")

mattfeiszli [12:54 PM] 
should origin be matt given my setup?

brian [12:54 PM] 
sorry, yes.

[12:55] 
good catch :simple_smile:

mattfeiszli [12:55 PM] 
no problem…  Peter always complains about my backwards setup.  :simple_smile:

brian [12:55 PM] 
crazily enough, your setup is the same as mine...and i'm still messing up :simple_smile:

mattfeiszli [12:55 PM] 
Ha!

[12:55] 
~/gitclients/PaperSDKCore (feature/shapeSampleRate3) $ git push -u matt feature/shapeSampleRate3
Branch feature/shapeSampleRate3 set up to track remote branch feature/shapeSampleRate3 from matt by rebasing.
Everything up-to-date

brian [12:55 PM] 
cool. now try `git pr o`

mattfeiszli [12:56 PM] 
(2-factor auth delay… but it appears to be working…)

[12:56] 
disco… that did it.

$ git remote -v
FiftyThree  [email protected]:FiftyThree/js2model.git (fetch)
FiftyThree  [email protected]:FiftyThree/js2model.git (push)
origin  [email protected]:FiftyThree/js2model (fetch)
origin  [email protected]:FiftyThree/js2model (push)

FiftyThree got added as part of a merge call

Fix up gemfile description

Barfs if target branch isn't mapped locally

First noticed by @dnkv when trying to merge into 'hotfix/2.0.2' branch