Dev Maintenance

Updating your code

Warning: VERY IMPORTANT: If you have custom changes that you have not committed or patched out, you want to do that first! If you follow these steps without doing so, you will likely muck things up.

It is best to shut your Apache instance down before doing the update process, to make sure that everything, especially scripts in cgi-bin/ are reloaded properly:

sudo /etc/init.d/apache2 stop

Okay. Let's say you've been running your Dreamwidth install and you want to pull down the latest and greatest in fixes. This is pretty easy.

cd $LJHOME
bin/cvsreport.pl -update
# only sync changes from the CVS to the live code
bin/cvsreport.pl -sync -cvsonly
bin/cvsreport.pl -sync -cvsonly

Yes, you do it twice. If the multicvs.conf file changed, then the first one pulls that file in, and the second one will notice whatever changed according to that file. 99% of the time once is enough, but every once in a while... so it's a good habit to have.

# order of commands is important
$LJHOME/bin/upgrading/update-db.pl -r -p --innodb
$LJHOME/bin/upgrading/update-db.pl -r --cluster=all --innodb
$LJHOME/bin/upgrading/texttool.pl load

Now you can restart Apache:

sudo /etc/init.d/apache2 start

Of course, in a production environment, this whole process is not too recommended as you never know what kind of code you're going to get. But for the most part, it's fairly straightforward. (And if you're doing development, this is generally safe.)

Updating to the production branch

You can see what branch you are currently on by going to $LJHOME/cvs/dw-free:

cd $LJHOME/cvs/dw-free
hg branch

Work is being done on the 'production' branch. When you create a new copy of the repository, it gets the branch with the label 'tip'. This should be the 'production' branch. You can sync to the 'tip' branch by using:

hg update -C tip

Warning: The -C option means "clean" and it removes all your local modifications.

If for some reason 'tip' is not pointing to 'production', then you can sync to 'production' manually by doing 'hg update -C production'

To be running production code, you also need to sync to your live code from the repository copy:

bin/cvsreport.pl -sync -cvsonly

Updating your system

On Ubuntu, you can update the packages on your system with:

apt-get update
apt-get upgrade

Or, if you want an easy command to run on your dw account in one swoop:

sudo apt-get update && sudo apt-get upgrade

Scripting

You can, of course, use scripts to make it easier for you to do most of this. Some example scripts are given below.

dwu - Updating the repos

Put this code in a file called ~/bin/dwu and make it executable with chmod ugo+x ~/bin/dwu:

#!/bin/bash
hg -R $LJHOME/cvs/dw-free qpop -a && \
$LJHOME/bin/cvsreport.pl -u && \
hg -R $LJHOME/cvs/dw-free qpush -a

Now, when you type 'dwu', this script will pop out any MQ patches you may have currently in dw-free, update the code to the latest available, and push your dw-free MQ patches back.

If you need support for dw-nonfree too, just duplicate the lines mentioning dw-free and edit them to say dw-nonfree and make sure that all the lines apart from the first and last end with '&& \':

#!/bin/bash
hg -R $LJHOME/cvs/dw-free qpop -a && \
hg -R $LJHOME/cvs/dw-nonfree qpop -a && \
$LJHOME/bin/cvsreport.pl -u && \
hg -R $LJHOME/cvs/dw-free qpush -a && \
hg -R $LJHOME/cvs/dw-nonfree qpush -a

dws - Syncing the live code

Put this code in a file called ~/bin/dws and make it executable with chmod ugo+x ~/bin/dws:

#!/bin/bash
$LJHOME/bin/cvsreport.pl -s -c && \
$LJHOME/bin/cvsreport.pl -s -c

This script will synchronise the live code twice when you type 'dws'.

dwdb - Updating the database

Put this code in a file called ~/bin/dwdb and make it executable with chmod ugo+x ~/bin/dwdb:

#!/bin/bash
$LJHOME/bin/upgrading/update-db.pl -r -p --innodb && \
$LJHOME/bin/upgrading/update-db.pl -r --cluster=all --innodb && \
$LJHOME/bin/upgrading/texttool.pl load

This will update the database when you type 'dwdb'.

tocvs - Switch between /cvs and live directories

Put this code in a file called ~/bin/tocvs.real, but do not make it executable:

#!/bin/bash
DIR=`pwd`
REPO=${1-'dw-free'}
 
if  [[ "$REPO" == "f" ]]; then REPO="dw-free";    fi
if  [[ "$REPO" == "n" ]]; then REPO="dw-nonfree"; fi
 
if [[ "$DIR" =~ ^$LJHOME ]]; then
  if [[ "$DIR" =~ ^$LJHOME/cvs/([^\/]+) ]]; then
    FROMREPO=${BASH_REMATCH[1]}
    if [[ "$1" == "" ]]; then
      DIR=${DIR/#$LJHOME\/cvs\/$FROMREPO/$LJHOME}
    else
      DIR=${DIR/#$LJHOME\/cvs\/$FROMREPO/$LJHOME\/cvs\/$REPO}
    fi
  else
    DIR=${DIR/#$LJHOME/$LJHOME\/cvs\/$REPO}
  fi
fi
cd $DIR

Now edit your ~/.bashrc file and add the following line to the end:

alias tocvs="source ~/bin/tocvs.real"

Then, log out and back in for the change to take effect. Now, whenever you're in a code directory, you can use 'tocvs' to switch back and forth from the live directory and the corresponding /cvs directory. It defaults to 'dw-free'; if you wish to go to another repository instead (for example, 'dw-nonfree'), give it as a parameter. (for example, 'tocvs dw-nonfree'). If you specify a repo name while you're already in a repository, it will switch to the named repository.

The repositories 'dw-free' and 'dw-nonfree' are aliased as 'f' and 'n' respectively.