If you're anything like me, you want to participate in the open-source community but don't want to screw things up. I invite you to participate in a safer side of open source.
In the past couple of articles (1 and 2), I've focused on higher-level community musings. I did this to encourage engagement and convey where engagement is already taking place. In short, open source on IBM i is growing significantly faster than it has in the past.
Now it's time to dive back into some geekier topics.
Before I get too much deeper, it would be good to set the stage for this article. Earlier this week, I had somebody ask whether they could compile RPG code on their Litmis Spaces account. Litmis Spaces accounts are given a browser-based editor for files in the IFS and are traditionally used for Node.js, Ruby, and Python (PHP is on the way). RPG source is no stranger to the IFS, as we learned in my "Shoot for Orion with Your RPG" article, but there was an additional complication in this situation because of how Litmis Spaces are locked down with chroot environments (a topic for another article). In short, I couldn't easily invoke /QSYS.LIB objects. You can see this unfold in this conversation I had with IBMer Tony Cairns.
In short, Tony wrote some code, and a new open-source project named db2util was formed—a PASE command that can communicate directly with DB2 for i. I think this utility will have many future uses, including automating IBM i tasks via shell scripts. But this article was promised to be geeky rather than musing, so let's dive into the makeup and compilation of db2util.
The db2util project is a combination of a setup shell script (cpysqlincludes.sh), a file named Makefile, and C code(db2util.c). This project is fairly simple on the dependency front, which makes it a great candidate for those desiring to compile their first C language IBM i OSS (open-source software) PASE project in the IFS.
Let's first look at the cpysqlincludes.sh file, as shown below.
ls /QIBM/include/sql* > ./list2
for i in $(< ./list2)
echo "PREPARATION COPY"
system -v "CPY OBJ('$i') TODIR('/usr/include/') TOCCSID(*STDASCII) DTAFMT(*TEXT) REPLACE(*YES)"
The first line has what's called a "shebang" and declares to the program loader what shell should be used to run this script. In this case, the full path to the Korn shell (ksh) was specified. The important thing to note is this project will be installed on many customer machines, and we should therefore focus on the least-common denominator. For example, I wouldn't want to put #!/QOpenSys/usr/bin/bash because not all shops would have the bash shell installed.
The next line obtains a list of all entries in /QIBM/include that start with sql. Why? Well, we need to have the SQL CLI API prototypes available to us in ASCII form so the db2util.c program can adequately resolve the #include statements. An interesting tidbit to note that wasn't immediately obvious to me when I started doing PASE things is that some symbolic links in the IFS/PASE environment point at things in the /QSYS.LIB environment, as shown with the below ls command.
$ ls -al /QIBM/include
. . .
lrwxrwxrwx 1 qsys 0 72 Oct 3 2015 sql.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQL.MBR
lrwxrwxrwx 1 qsys 0 76 Oct 3 2015 sqlca.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLCA.MBR
lrwxrwxrwx 1 qsys 0 78 Oct 3 2015 sqlcli.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLCLI.MBR
lrwxrwxrwx 1 qsys 0 76 Oct 3 2015 sqlda.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLDA.MBR
lrwxrwxrwx 1 qsys 0 78 Oct 3 2015 sqlenv.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLENV.MBR
lrwxrwxrwx 1 qsys 0 76 Oct 3 2015 sqlfp.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLFP.MBR
lrwxrwxrwx 1 qsys 0 80 Oct 3 2015 sqlscds.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLSCDS.MBR
lrwxrwxrwx 1 qsys 0 82 Oct 3 2015 sqlstate.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLSTATE.MBR
lrwxrwxrwx 1 qsys 0 82 Oct 3 2015 sqlsystm.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLSYSTM.MBR
lrwxrwxrwx 1 qsys 0 78 Oct 3 2015 sqludf.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SQLUDF.MBR
lrwxrwxrwx 1 qsys 0 72 Oct 3 2015 ssl.h -> /QSYS.LIB/QSYSINC.LIB/H.FILE/SSL.MBR
. . .
A PASE symbolic link is like a Windows desktop shortcut; it's a conveniently located shortcut to a resource in another location. As we can see above, there are a number of /QIBM/include files that link to /QSYS.LIB/QSYSINC.LIB/H.FILE.
Next up is the file named Makefile. Think of a Makefile as being analogous to a CL program used to consolidate RPG compile commands: Variables are set, compile statements like CRTBNDRPG are run, etc. I won't copy the entire contents of the Makefile here; instead I’ll describe some of its content and purpose. A Makefile is processed by the make command utility. The make command utility is used to compile projects in incremental fashion. It does this by comparing timestamps of source files and compiled files. If the source file has a timestamp that's more recent than the compiled file, it knows a recompile is in order. Inside of the Makefile, you'll find a variety of aspects similar to the aforementioned CL program, including the setting of variables (e.g., CC=gcc) and the eventual compilation of the targeted item(s), in this case the below command. It's worth noting the contents of a Makefile is in large part shell commands. In short, start learning shell scripting.
$(CC) $(CCFLAGS) $(DB2UTILLIBOBJS) $(DB2UTILLIBDEPS) -o $(DB2UTILPGM32)
The above line may initially look complicated, but it's actually simple. The $(...) syntax is evaluating the contents between the open and close parentheses. In the case of $(CC), it is simply returning the value contained in CC, which in this case is "gcc". Then it moves to $(CCFLAGS), which was populated from variable CCFLAGS32, which contains the value of "-g", an option for the gcc compiler command. This continues until we have a command similar to the following:
gcc -g db2util.o -L. -lpthreads -liconv -ldl -lpthread -ldb400 -o db2util
Think of this as being similar to programatically composing a CRTBNDRPG compile command based on a variety of situations. For example, maybe you compile the RPG differently based on what version of IBM i is discovered.
To kick off the processing of the Makefile, you simple run the following commands:
$ cd /path/to/project/code
Running the make command will search for a Makefile in the current directory and start processing it.
OK, so up to this point I've described the project and how to compile it, but I haven't told you how to obtain the project or how to obtain the gcc compiler. There are two approaches to take. The first is to use Git to clone the repositories right onto your IBM i. If you don't have Git, then download the db2util repository and FTP it to the IFS. Below, I show the Git clone approach.
First we will obtain the gcc compiler by cloning the IBM i chroot project. This is also shipped as 5733OPS option 3.
$ cd /QOpenSys
$ git clone firstname.lastname@example.org:litmis/ibmichroot.git
$ cd ibmichroot
Now run the pkg_setup.sh shell script and specify the pkg_perzl_gcc-4.8.3.lst file, which contains URLs to all the necessary dependencies for gcc to work, along with gcc itself.
$ ./pkg_setup.sh pkg_perzl_gcc-4.8.3.lst
When you run pkg_setup.sh you'll see many messages like the following. I've only included snippets of the actual output because there are many hundreds of lines.
failed to stat /QNTC: No such file or directory
group system does not exist - using root
Processing file: pkg_perzl_perl-5.8.8.lst
Resolving www.oss4aix.org... 184.108.40.206
Connecting to www.oss4aix.org[220.127.116.11]:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 25,200,295 [application/x-rpm]
Once the script is complete, you can check to see if gcc was successfully installed using the below commands.
$ which gcc
$ gcc --version
gcc (GCC) 4.8.3
Copyright (C) 2013 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
Yahoo! The gcc compiler was successfully installed. Next, use Git to clone the db2util project.
$ cd /QOpenSys
$ git clone email@example.com:litmis/db2util.git
$ cd db2util
At this point, we're back to using the make command to compile the project. Once the make command completes, you can go back to the db2util project home page to learn about usage scenarios and examples.
As you might guess, I've only skimmed the surface of what make and gcc are capable of. My goal in this article was to give an introductory exposure to "how things work." Let me know what other things you'd like expounded upon. Please comment below or email me at firstname.lastname@example.org with any questions you have.
About the author: Aaron Bartell