pfg-presentation

github ethanj 5El5R5IBfN1s


# install hub for extra git commands for github
# https://hub.github.com/hub.1.html
brew install hub && echo "\n#add github commands to git\neval \"\$(hub alias -s)\"" >> ~/.zshrc

# create a new private repo and setup remote
git create -p philosophie/foo
git remote add origin git@github.com:philosophie/foo.git
git push -u origin master

---

aliases in ~/.gitconfig:
[alias]
  co = checkout
  ci = commit
  st = status
  br = branch
  hist = log --pretty=format:\"%h %ad | %s%d [%an]\" --graph --date=short
  type = cat-file -t
  dump = cat-file -p

----

#cache passwords
git config --global credential.helper cache
#adds the following to .gitconfig:
#[credential]
#	helper = cache

#nice reference
http://githowto.com/

#dont forget about the gui interface... gitk
gk

-----


#start a new repo from existing directory
git init .

git add .

git commit -a -m "initial commit"

git remote add origin git@github.com:philosophie/dkind-auto-frame.git
git push origin master

#get the most recently committed version of a file
git checkout foo.html

#get a previous version of a file
#the following gets lesson.js 2 commits ago
git show HEAD~2:html/js/lesson.js > foo
#relative to current dir rather than root
git show HEAD~2:./lesson.js > foo

#checkout to a previous version in the working directory
#make sure current version is committed
git commit -a -m "checkpoint"
#list previous versions
git hist
#switch to one
go xxxx
#switch back to current
go master


#create new local branch
git checkout -b <branchName>

#delete local branch (merged and unmerged)
git branch -d <branchName>
git branch -D <branchName>

#delete remote branch
git push origin --delete <branchName>


#show just committed diff
git diff HEAD^..HEAD
#just the files:
git log --name-status HEAD^..HEAD
#just the comment
git log -1



#list current repo state of files in current directory
#equiv to svn list .
git ls-tree --name-only HEAD

#revert local changes back to previous commit (or any version), deleting all uncommitted changes locally (working dir matches repo)
#note the data is still in git somewhere (git hist --all)
git reset --hard HEAD

#remove directory
git rm -r xxx/


#remove changes from the staged commit, but keep the changes in working directory
git reset HEAD hello.html

#cancel the last commit
git revert HEAD


#IMPORTANT
#undo the last commit, keeping the changes, assuming you did not push the changes publicly yet
#if you did push the changes, you can still do it, but will need to force push to get it updated correctly
git reset HEAD~1 --soft

for instance, say you commit to master instead of a branch, undo
it using soft, switch to a branch and commit

# undo the last push to master, throwing out the commit
# WARNING: once pushed, this erases all trace from the repo and cant be recovered
git reset --hard HEAD~1
git push -f
#or
git push -f origin <sha_of_previous_commit>:master


####################################

#working with a team and pull requests

#make sure in the master branch
go master

#do some work...

#create a new branch
#git co -b ejbranch
go -b ejbranch
#commit your work
#git commit -a -m "comment"
gca "comment"

git commit -m "BACKOFFICE-XXX #time 8h Work done

#upload the work to remote repo
#git push origin ejbranch
gpush ejbranch

#specific to java in eclipse:
#before doing a pull request, run maven from commandline to make sure everything compiles correctly
mvn clean install
#then do a project/clean in eclipse for all to remove maven artifacts that are different from eclipse
project/clean

#may need to rebase?
[8:49:28 AM] Kristian: git fetch -p
[8:49:37 AM] Kristian: git rebase origin/master
[8:49:42 AM] Kristian: git commit
[8:49:47 AM] Kristian: then git push

#create a pull request with no reviewers
#review all changes in the PR to make sure happy with it
#then add reviewers to the PR

#wait for the pull request to be merged
#while waiting you can work off master branch again
go master
git pull


#delete the local branch
#git branch -d ejbranch
gb -d ejbranch


#repeat the cycle


---------


#bring working directory up to date with master
#if still in master:
git stash
go master
git stash apply


#resolve conflict in a pull request:
#first merge master branch changes locally
go master
git pull
go ejbranch
git merge master
#edit the files that have conflict to get rid of them
gca "fixed conflicts"
gpush ejbranch


------------------------

#IMPORTANT
#for when you commit to master before switching to a new branch, but did not push it yet:
#undo the last commit, keeping the changes, assuming  you did not push the changes publicly yet
git reset HEAD~1 --soft

#create the new branch, etc


###############################################

#useful.. follow this with pull to deploy live
git commit -a -m "blah"; git push ec2

#remove untracked files
git clean -n
git clean

#add empty dirs to the repo, like log dirs
#create a .gitignore inside that directory that contains four lines:
#note only do this in the leaf dirs
.gitignore:

# Ignore everything in this directory
*
# Except this file
!.gitignore

then:
git add dir

#this should also work, but will complain about untracked files presumably
find . -type d -empty -exec touch {}/.gitignore \;

###############################################


#setup git to deploy click project remotely
#as ec2-user
mkdir /web /web/click
mkdir ~/repos ~/repos/click
cd /home/ec2-user/repos/click
git init --bare
cd /web/click
git clone /home/ec2-user/repos/click .
#now you can do git pulls in the production dir
cd /home/repos/click/hooks
vi post-receive
#( add post-receive hook content, make it executable)
#You can add a remote to the code repo with
git remote add prod /web/click
#With that setup, you can have your post-receive hook do git push prod master.

#allow automated push live
cd ~/repos/click
git config --global push.default simple
cd /web/click
git config receive.denyCurrentBranch ignore


#locally...
git remote add ec2 ec2-user@clickredir:/home/ec2-user/repos/click
git push ec2 master

#commit, push, deploy and reload apache in one line
git commit -a -m "blah"; git push ec2; ssh -t ec2-user@clickredir.com 'cd /web/click; git pull; sudo /etc/init.d/apache reload'



###############

#change ips on aws servers
git remote set-url web1 ec2-user@ec2-54-159-100-164.compute-1.amazonaws.com:/home/ec2-user/repos/click
git remote set-url web2 ec2-user@ec2-54-224-93-37.compute-1.amazonaws.com:/home/ec2-user/repos/click
git remote set-url web3 ec2-user@ec2-54-162-190-91.compute-1.amazonaws.com:/home/ec2-user/repos/click



###############################################


#setup git to deploy cpc project remotely
#as ec2-user
mkdir /web /web/cpc
mkdir ~/repos ~/repos/cpc
cd /home/ec2-user/repos/cpc
git init --bare
cd /web/cpc
git clone /home/ec2-user/repos/cpc .
#now you can do git pulls in the production dir


#push the initial project state
#locally...
git remote add <remote name, ie web1> ec2-user@<new amazon server>:/home/ec2-user/repos/cpc
git push <remote name> master


#EDJ - don't do this next as current process is to cd to the production dir and pull via the clone setup
#cd /home/ec2-user/repos/cpc/hooks
#vi post-receive
##( add post-receive hook content, make it executable)
##You can add a remote to the code repo with
#git remote add prod /web/cpc
##With that setup, you can have your post-receive hook do git push prod master.

#allow automated push live
#not needed if already configured for previous project
#cd ~/repos/cpc
#git config --global push.default simple
#cd /web/cpc
#git config receive.denyCurrentBranch ignore

#--------------------------------

#add a new web server ec2 instance

ami: click_server_3
us-east-1a
basic monitoring

#locally...
git remote add <remote name, ie web1> ec2-user@<new amazon server>:/home/ec2-user/repos/click
git push web1;
ssh -t ec2-user@<new amazon server> 'cd /web/click; git pull; sudo /etc/init.d/apache reload'

#make sure couchbase not running, and that the moxi instance that is running is connect to moxi servers rather than locally

#test direct in browser
http://ec2-184-73-29-172.compute-1.amazonaws.com/index.html?g=2&debug=prn1!

#add to load balancer



###############################################


#setup git to deploy feed project remotely
#as ec2-user
mkdir /web /web/feed
mkdir ~/repos ~/repos/feed
cd /home/ec2-user/repos/feed
git init --bare
cd /web/feed
git clone /home/ec2-user/repos/feed .
#now you can do git pulls in the production dir


#push the initial project state
#locally...
git remote add <remote name, ie web1> ec2-user@<new amazon server>:/home/ec2-user/repos/feed
git push <remote name> master


###############################################


#mount one directory to another instead of symlinks so git tracks it
#http://stackoverflow.com/questions/86402/how-can-i-get-git-to-follow-symlinks

#NOTE these are all done automatically in /etc/fstab, ie:
#/root/Dropbox/k8corp/release /web/ccp/html/assets/k8 none bind 0 0

mount --bind /root/Dropbox/k8corp/release /web/ccp/html/assets/k8
mount --bind /root/Dropbox/k8corp/ethan /web/ccp/html/assets/dev/k8

#share modules dir between projects but let git track
mount --bind /web/prn/modules /web/cpc/modules
mount --bind /web/prn/modules /web/feed/modules
mount --bind /web/prn/modules /web/video/modules









localhost:/web/cpc# chmod -R +x bin/
localhost:/web/cpc# chmod -R +x cgi-bin/
localhost:/web/cpc# chmod +x cgi-bin/*
localhost:/web/cpc# chmod +x bin/*
localhost:/web/cpc# chmod +x bin/ZohoReports/UploadTool/jre/bin/*
localhost:/web/cpc# chmod +x bin/ZohoReports/UploadTool/bin/*
localhost:/web/cpc# ls -lbin/ZohoReports/UploadTool/bin/C^C
localhost:/web/cpc# ls -l bin/ZohoReports/UploadTool/bin/C
ls: cannot access bin/ZohoReports/UploadTool/bin/C: No such file or directory
localhost:/web/cpc# ls -l bin/ZohoReports/UploadTool/bin/
total 32
-rwxr-xr-x. 1 root root 1633 Nov 21  2013 CSVUploadConsole.sh*
-rwxr-xr-x. 1 root root  116 Nov 21  2013 CSVUploadGUI.sh*
-rwxr-xr-x. 1 root root  478 Nov 21  2013 setEnv.sh*
-rwxr-xr-x. 1 root root 1377 Nov 21  2013 UploadFromDB.sh*
localhost:/web/cpc# chmod -x bin/ZohoReports/UploadTool/bin/*
localhost:/web/cpc# chmod +x cgi-bin/admin/editCpcConfig
localhost:/web/cpc# chmod +x bin/unused/aggregateCounts
localhost:/web/cpc# cd etc
localhost:/web/cpc/etc# cd ..
localhost:/web/cpc# cd etc
localhost:/web/cpc/etc# chmod +x moxi-cpc
moxi-cpc        moxi-cpc-local
localhost:/web/cpc/etc# chmod +x moxi-cpc* zohocron
zohocron       zohocronDaily  zohocronRT
localhost:/web/cpc/etc# chmod +x moxi-cpc* zohocron*
localhost:/web/cpc/etc# chmod +-x startupCpc.pl


#####################################################


#squash commits of a branch
#merge the commits of a branch into one to simplify pull requests, etc

git rebase -i HEAD~7

#where 7 is the number of commits in the branch
#it will show a list of the commits
#leave the first line with pick ...
#for the other lines
#you will replace pick with 's'
#and then go ahead and
:x
#to save and proceed
#it should show you how the comments for your squashed commit should look
#you can edit it as you like
#and then save it
:x
#now you should be in the command line again
#it should show you something like Successfully rebased and updated
#now force a push
git push -f


#####################################################

philosophie

#github repo comment for approved
:shipit


----

#for a branch named feature/prompts

git branch feature/prompts
aka
go -b feature/xxx
gb feature/xxx


#do some work
#do some commits
gca ....
gpush feature/prompts
gca ...
gpush feature/prompts
...

#when ready to merge
#push branch to github
gpush feature/prompts

#make a pull request on github
#get the shipit squirrel

#get the last commit where my branch diverged from master
git merge-base feature/prompts origin/master
> ea3f7d7159bc31a7855418982ee9f6858355feac
#squash commits and edit the commit message
git rebase -i ea3f7d7159bc31a7855418982ee9f6858355feac

#pushed the squashed version into the branch as a replacement
git push -f origin feature/prompts

#check github for merge conflicts

#go back to master and merge in the branch
go master
git pull
git merge feature/prompts
gpush master
#delete branch in github
#delete local branch
git branch -d feature/prompts


##############

rebase example

go -b feature/sponsors
gca "make sponsors database driven, with images on s3"
gpush feature/sponsors
#post a PR
#do some work inteneded for a new branch
git stash
go master
go -b feature/signup
git stash apply
gca "implement form submission for signups"
gpush feature/signup
#make edits to sponsors PR
go feature/sponsors
gca "more adjustments for PR"
gpush
#sponsors PR done
go master
gb -d feature/sponsors
git pull
#switch back to new branch and rebase off new state of master
go feature/signup
git rebase master
#force push is required
gpush -f


##############

#switch to someone else's branch to test stuff
git fetch
go branch



#################

#get a diff since the branch was created

#get the last commit where the branch diverged from master
git merge-base feature/prompts origin/master
> ea3f7d7159bc31a7855418982ee9f6858355feac

git diff ea3f7d7159bc31a7855418982ee9f6858355feac


#############

#keep folder in git but not contents
put this .gitignore into the folder, then git add .gitignore

*
*/
!.gitignore


#################

#work with a fork of a repo

#grab new version of master from the original repo
git remote add upstream git@github.com:uport-project/muport-core-js.git
git fetch upstream
git rebase upstream/master
git push origin master

#if there is an error about conflict and doesnt seem to resolve (because nothing changed), do
git rebase --skip
#which skips over just that conflict

#################

#pull in new master stuff while in a different branch

#make sure your branch is committed
gcam "lalla"

#update local master
gco master
git pull

#switch back the branch and rebase
gco branch
git rebase master

#this will replay your local branch on top of the new local master


####################

# see what files changed in a repo between 2 recent commit points
git diff --name-only HEAD~10 HEAD~5


#push an expty commit, ie to force a deploy with CI
git commit --allow-empty
git push ...

####################

# work on new dependent branch (newbranch) while waiting to merge pr on dependent branch (prbranch)

git checkout -b newbranch
# equiv to following if not on prbranch
git checkout -b newbranch prbranch

#do work on newbranch

# lets say pr needs mods
gco prbranch
# do edits
# commit changes to prbranch
gcam "cleanup for pr"
git push origin prbranch
# incorporate changes to prbranch into newbranch
gco newbranch
git rebase prbranch
# resolve conflicts if present (git add + git rebase --continue)

# pr gets approved
# squash and merge prbranch in github

# important non-obvious way to "rebase" newbranch to result of prbranches merge
# on newbranch, squash newbranch commits into one commit
git log
# search for the commit right before newbranch commit
# this should be labelled with the text of the last commit of the squashed PR on github
git rebase -i <sha of previous commit>
# change first letter of each line to 's' except for the first line!
# retrieve sha of the resulting squashed commit for use later, aka 'newsha'
# which is the firs commit/sha in the output of git log
git log

gco master
git pull
# rename newbranch to a temp name, since we will create a new branch based on it
git branch -m <newbranch> tmpbranch
# recreate our new branch
gco -b <newbranch>
# cherry pick the squashed commit from tmpbranch to newbranch
git cherry-pick <newsha>
# force push newbranch to github otherwise it will reject it
git push origin <newbranch> -f
# newbranch is now "normal" and has original name

# pr gets merged
# Rebase new branch onto master
git pull --rebase origin master



########################

# create a gist from clipboard with short url copied to clipboard afterwards
gist --paste --shorten --copy --filename "blah"