Caelum's Blog

The freely available Windows SDK (current version) contains everything you need including a set of compilers. If however you are getting the Windows 8 SDK you will also need to get Visual Studio Express 11 for the compilers, as the Windows 8 SDK no longer includes a set of compilers.

Now you need to modify your environment to make the SDK tools available to perl (CPAN installations, etc..) I recommend using Rapid Environment Editor, or you can make .cmd file and modify your cmd.exe shortcut to add the /K option and the path to your .cmd file with your settings. I attempted, with some measure of success, to make a system of switching between different perls in a single cmd.exe session in this article.

The settings you need for the Windows 7 SDK are as follows (modify to taste:)

set PATH=C:\Perl64\site\bin;C:\Perl64\bin;C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\Bin\amd64;C:\Program Files\Microsoft SDKs\Windows\v7.1\Bin\x64;C:\Windows\system32;C:\Windows
set INCLUDE=C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\INCLUDE;C:\Program Files\Microsoft SDKs\Windows\v7.1\Include;C:\Program Files\Microsoft SDKs\Windows\v7.1\Include\gl
set LIB=C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\Lib\amd64;C:\Program Files\Microsoft SDKs\Windows\v7.1\Lib\X64
set LIBPATH=C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\Lib\amd64

Now launch cpan in cmd.exe and try to install an XS module such as Class::XSAccessor to test everything out.

If you get link errors about not being able to find bufferoverflowU.lib, edit C:\Perl64\lib\Config_heavy.pl and remove all instances of bufferoverflowU.lib. This shouldn't happen with the newer ActivePerl builds however.

The settings for the Windows 8 SDK with the Visual Studio Express 11 compilers should be similar, but I have not tried that yet, so don't know the details.

The ActivePerl MinGW mode mostly works with MinGW-w64, although it is meant for the 32bit builds with older mingw.org suites so some tweaks are necessary.

First download the WIN64 native compiler suite from the MinGW-w64 download page in the WIN64 Downloads -> Personal Builds -> rubenvb -> release section. The file will be named something like x86_64-w64-mingw32-gcc-4.7.0-release-win64_rubenvb.7z. Unpack with 7-zip or WinRAR everything below the top level directory in the archive to C:\MinGW_w64 .

Modify your PATH to add C:\MinGW_w64\bin . See the part about environment settings in the previous section, I recommend Rapid Environment Editor.

Set the LIBRARY_PATH environment variable as follows:

set LIBRARY_PATH=C:\MinGW_w64\x86_64-w64-mingw32\lib

Now install dmake. We'll use a 32bit binary from CPAN so make sure that WOW64 is enabled in your OS if you are using an older server OS. Download the dmake dist from CPAN, open the archive, go into the dmake folder, and extract dmake.exe and the startup subfolder into C:\Perl64\site\bin . Then run:

icacls c:\perl64\site\bin\dmake.exe /grant your_user_name:(rx)

Next edit C:\Perl64\lib\Config_heavy.pl and remove all instances of bufferoverflowU.lib.

That's all there is to it. Now with gcc.exe in your PATH ActivePerl will go into mingw mode when it needs to build an extension, try it now by running cpan Class::XSAccessor or some such extension.

Note that if you will package Perl applications with built extensions, you will need to include the libgcc and libstdc++ (if there were any C++ extensions) DLLs with your package, to be run from the same directory as the perl interpreter or somewhere in PATH. This is also true for the new versions of the 32bit mingw.org toolchain. The alternative is to include

-static-libgcc -static-libstdc++

in your LDDLFLAGS along with the default $Config{lddlflags}. I'm going to submit a bug report to ActiveState to make these options default for newer gccs.

Plug in your USB/ESATA/FireWire hard drive into your linux computer, then do a:

dmesg | tail

to note the block device, such as /dev/sdd.

Next you will want to partition it. Use cfdisk or gparted.

Make one primary partition of type Linux (83) that is the size of the drive.

Next format it with your favorite file system.

Use a command such as:

sudo mkfs.ext4 /dev/sdd1

Now let's mount it. We are going to mount it under /media so that the updatedb cron job does not index it and clutter your locate db with your backup files. We are also going to mount it by UUID so that the order you plug in your devices, the presence of hubs, etc. does not affect mounting.

First make the mount point:

sudo mkdir /media/backup

Then get the UUID of the new partition:

sudo blkid /dev/sdd1

Now add the mount point to /etc/fstab using the UUID from the previous step:

UUID=e3b623e7-2049-46cf-89aa-1152c38e08e8 /media/backup ext4 defaults,nofail

Now mount it!

sudo mount /media/backup

Remember to mount it when you plug your hard drive in.

If you know of a way to automount an fstab entry, please email me so I can add it here...

We are going to make the backup directory one level below the mount point so that having the drive disconnected tells Back In Time to not run the backup job, otherwise it would run the backup job on the root file system!

sudo mkdir /media/backup/snapshots

Now you're ready to run the Back In Time configuration program. Run:

ssh -X root@localhost backintime-gnome

Click the Settings button.

For "Where to save snapshots" select /media/backup/snapshots.

For "Schedule" I use "Every Day" at 03:00. Choose a time your machine is likely to be idle and the hard drive is likely to be plugged in.

Click the Include tab. Add folders such as /home and /etc and any others you want to back up.

Press "Ok" and close the program.

Press Ctrl-C to close the ssh session.

And you're done!

I also highly recommend installing anacron:

sudo aptitude install anacron

especially on a laptop, because Back In Time runs from cron. This way if your laptop is off or asleep when the job is supposed to run, your backup will still run at some future time.

Backups can seriously save your arse. Hard drives and SSDs fail. Alot.

Now Mac OS X and Windows have attractive built-in backup options, Time Machine and Windows Backup, respectively. Time Machine is unmatched in power and simplicity. Windows Backup is not as shiny, but still does the job admirably.

With Back In Time, we now have a nice backup solution for Linux as well.

Enjoy the comfort of knowing that your data is safe.

It is available here: http://www.sybase.com/detail?id=1016644.

Open the package and double click the Install SQL Anywhere icon, select all components.

At the end click the Check for Updates button, it will take you to the EBF download. You will need to create a MySybase account to download the EBF package. Download the package, doubleclick the Install SQL Anywhere icon and select the update product option.

Add the following to your ~/.bashrc:

export SQLANYWHERE_HOME=/Applications/SQLAnywhere12/System
export DYLD_LIBRARY_PATH=$SQLANYWHERE_HOME/lib64:$SQLANYWHERE_HOME/lib32:$DYLD_LIBRARY_PATH
export DYLD_BIND_AT_LAUNCH=1
export PATH=$PATH:$SQLANYWHERE_HOME/bin64:$SQLANYWHERE_HOME/bin32

For the next step, open a new terminal session for the environment changes to take effect.

Execute the following to create a new database:

cd
mkdir sqlanywhere
cd sqlanywhere
mkdir logs
dbinit -t ./logs/hlaghdb.log -p 4096 -dba DBA,SQL hlaghdb.db

Create a file called ~/Library/LaunchAgents/com.ianywhere.sqlanywhere.plist, put the following in that file, changing /Users/rkitover to your actual home directory and the database file name to your actual database file:

<?xml version='1.0' encoding='UTF-8'?>
<!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN"
"http://www.apple.com/DTDs/PropertyList-1.0.dtd" >
<plist version='1.0'>
<dict>
<key>Label</key><string>com.ianywhere.sqlanywhere</string>
<key>WorkingDirectory</key><string>/Users/rkitover/sqlanywhere</string>
<key>EnvironmentVariables</key>
    <dict>
        <key>DYLD_LIBRARY_PATH</key>
        <string>/Applications/SQLAnywhere12/System/lib64:/Applications/SQLAnywhere12/System/lib32</string>
        <key>DYLD_BIND_AT_LAUNCH</key>
        <string>1</string>
    </dict>
<key>ProgramArguments</key>
<array>
        <string>/Applications/SQLAnywhere12/System/bin32/dbsrv12</string>
        <string>-ud</string>
        <string>/Users/rkitover/sqlanywhere/hlaghdb.db</string>
</array>
<key>Debug</key><false/>
<key>Disabled</key><true/>
<key>KeepAlive</key><true/>
</dict>
</plist>

NOTE: we do not give the database server a name in this example. This is because for some reason when I tried to use -n option to dbsrv12 it responded with Communication error. Try running dbsrv12 from the command line with the -n option to see if it works for you, if it does add it to the ProgramArguments in the .plist file. E.g.:

dbsrv12 -ud -n SQLANYWHERE hlaghdb.db

If this works, add the following to the .plist file after dbsrv12:

<string>-n</string>
<string>SQLANYWHERE</string>

The following examples will all assume the database server has no name.

Now execute the following to load the service and start the database server:

launchctl load -w ~/Library/LaunchAgents/com.ianywhere.sqlanywhere.plist
launchctl start com.ianywhere.sqlanywhere

You should see the database server in ps:

$ ps aux | grep dbsrv
rkitover       51654   0.3  0.2  1627072  14000   ??  Ss    1:02PM   0:03.07 /Applications/SQLAnywhere12/System/bin32/dbsrv12 -ud /Users/rkitover/sqlanywhere/hlaghdb.db

It will be started across reboots.

launchctl stop doesn't work with this plist, so if you need to restart the server, just kill the process and do a launchctl stop followed by a launchctl start.

Open the SQLAnywhere12 folder under Applications in Finder.

The applications we're interested in are Sybase Central and Interactive SQL.

The Sybase Central application will allow you to configure all aspects of your database. To use it, open the application, double click on the SQL Anywhere 12 plugin, right click inside the blank window and choose Connect, choose Connect to a running database on this computer, enter the username and password you used with the dbinit command, which in this example is DBA and SQL respectively.

The Interactive SQL application is a handy GUI for testing SQL statements on your database. To login, open the application, choose Database authentication, choose Connect to a running database on this computer, enter your username and password and press Connect. Inside the application, enter your SQL statements in the SQL statements window and press F5 to execute them; you will see your results in the results window.

If you would prefer a commandline utility to Interactive SQL for testing SQL statements, you can use the sqsh utility from MacPorts (http://www.macports.org/.)

The sqsh utility is an interactive SQL utility for databases using the TDS protocol, such as Sybase ASE and Microsoft SQL Server. SQL Anywhere also uses the TDS protocol, so the utility works for these databases as well.

If you do not yet have an interfaces file, follow these steps to set one up.

In your .bashrc add the following:

export SYBASE=~/sybase

In this directory create a file called interfaces with the following:

SQLANYWHERE
  master tcp ether localhost 2638
  query tcp ether localhost 2638

Now you can launch sqsh like so:

sqsh -S SQLANYWHERE -U DBA -P SQL

To use it, type in a SQL statement, it can span multiple lines, then type go on a line by itself to execute it. This may take some getting used to if you're used to tools such as sqlplus or dbish, where SQL statements are terminated by a semicolon. You can use dbish with SQL Anywhere as well, I will describe how to do so further down.

If you have UnixODBC from MacPorts (http://www.macports.org/,) you can set up the included ODBC drivers with it.

In /opt/local/etc/odbcinst.ini add the following:

[SQLAnywhere]
Description = SQL Anywhere
Driver = /Applications/SQLAnywhere12/System/lib64/libdbodbc12.dylib
Setup = /Applications/SQLAnywhere12/System/lib64/libdbodbc12.dylib
FileUsage = 1

Now lets test it out (take out the -S flag to cpanm if you use local::lib):

curl -L http://cpanmin.us | perl - -S App::cpanminus
cpanm -n -S DBI
cpanm -n -S DBD::ODBC
perl -MDBI -le 'my $dbh = DBI->connect("dbi:ODBC:driver=SQLAnywhere", "DBA", "SQL"); print for $dbh->selectrow_array("select 42")'

First install DBI:

curl -L http://cpanmin.us | perl - -S App::cpanminus
cpanm -n -S DBI

Then build and install the Perl library:

cp -a /Applications/SQLAnywhere12/sdk/perl /tmp/sqlany
cd /tmp/sqlany
perl Makefile.PL
make
sudo make install
cd
rm -rf /tmp/sqlany

Test it out:

perl -MDBI -le 'my $dbh = DBI->connect("dbi:SQLAnywhere:", "DBA", "SQL"); print for $dbh->selectrow_array("select 42")'

If you don't like Interactive SQL or sqsh, you can try the Perl module DBI::Shell for a commandline SQL utility.

First install it:

curl -L http://cpanmin.us | perl - -S App::cpanminus
cpanm -n -S DBI::Shell

Then launch it. I would recommend using the ODBC driver, especially if you use UUID columns:

dbish dbi:ODBC:driver=SQLAnywhere DBA SQL

To use the Perl binding:

dbish dbi:SQLAnywhere: DBA SQL

SQL statements are terminated by semicolons, type /quit to exit the shell and /help to see what commands are available.

If you would like to save the readline history of your sessions and use it for completion, you can wrap it with rlwrap. To do so add this alias to your ~/.bashrc:

alias sqlany="rlwrap -a -N -t dumb -i -f ~/.sqlanywhere_dbish_history -H ~/.sqlanywhere_dbish_history -s 30000 dbish dbi:ODBC:driver=SQLAnywhere DBA SQL"

and initialize the history file:

touch ~/.sqlanywhere_dbish_history

then you can type sqlany to launch the shell.

There is complete support for SQL Anywhere in DBIx::Class and DBIx::Class::Schema::Loader (the Catalyst create=static helper.)

Hope you found this guide helpful. Feel free to leave comments on the blog or email me if you have questions.