Quick guide to Multi-master Replication in PostgreSQL

A while ago, I wrote a post about multi-master replication using symmetricDS. My scenario consists of a system with multiple nodes, all of them writing in their copies of the database. Sometimes the nodes may be offline, but I would like the system to be *eventually consistent*.

SymmetricDS is a Java-based framework that supports a number of RDBMS, including PostgreSQL, the one that I use. I don’t particularly like the fact that it uses Java: specially the UI, seems slow and unresponsive. However, the fact that the application is cross-platform is quite “handy”, as we can have the databases running in a number of different, and “talking to each other”.
SymmetricDS itself is free and Open Source (GPL). However, if you want to use the configuration GUI, there is a commercial product called “SymmetricDS Pro”. I could not find out how much the pro version costs (they are quite secretive in the website), but since I was in a bit of a rush to setup the synchronization system, I decided to try it out.
Previously, I evaluated the FOSS version, and was able to synchronize 2 databases on Ubuntu systems: what they called a “Standard 2 Tier Configuration”. This time, I went for a slightly more complicated scenario: synchronizing three different databases, all in different hosts, with a mix of Windows and Linux systems. With the help of the “pro” GUI, and the “Quick-start manual”, it took me less than two days to do it, which I think is ok.

Before start reading this post, please note that database replication is a complicated issue. Multi-master asynchronous replication is *definitely* complicated, with many things involved, so don’t expect the configuration to be a simple wizard. To be able to use it you need to understand well a series of concepts, that won’t take you five minutes. Having said this, “SymmetricDS Pro” does a pretty good job in helping a person that *has this concepts*, performing that task.

My case study, is a real world scenario where I have three different hosts running copies of my application and database. However it may be over-simplified, since I am doing simple operations with the application (inserting/updating data with all the nodes online). Asynchronous multi-master replication “gives space” for the rise of conflicts, and although SymmetricDS does provide some support for dealing with conflicts, this is a highly sensitive topic, that must be dealt on a “case-to-case” basis, by a person with a good knowledge of the domain. On my case study, I did not arrive to any conflicts so I won’t evaluate how symmetriCDS deals with them. Please have this issue in mind, if you decide to adopt SymmetricDS.

SymmetricDS Pro is not free, but you may download it and evaluate it for 30 days:


It is essential to give your email address, where they will provide you with the key to “unlock” the full functionality. I found it very easy to install it, following the instructions on the quick-start guide:


The only dependency is the Java Runtime Environment (JRE), which very likely you will already have running on your system, anyway.
In the guide they mention a “single-homed” scenario, where you will have a single instance of symmetricDS running and a “multi-homed” scenario, where you install a copy of symmetricDS for each host/database. Since I wanted to approach a “deployment scenario” with remote computers I went straight to the “multi-homed”. However, if you just want to test it, you may try the “single-homed” scenario (which is supported in the manual).

Although SymmetricDS enables a distributed system, you need to create a node that acts as a “registration server”. This node has to exist, even if you can make the other nodes “talk” to each other. Although it is ok if this node is offline for a while, I would pick a host that is mostly online (like a actual server).

I started by installing symmetriCDS in my “server” node. The installation is exactly the same on any node and when you finish, you start running the daemon (running something like “/symmetricDS/bin/sym”), and then run the node setup.

If you have installed symmetricDS on port 31415 (the default non-secure port), the configuration console can be run from pointing your browser to this address:


Since I was on the server host, I choose to setup a “server” node. SymmetricDS presents you with two “ready made” configurations, and an option to create your own, called: “I’ll configure things myself”. This is actually a very important step of your configuration, since it will define the architecture of the system (how many nodes you have, how they connect to each other, etc); later you may refine the configuration options, but the first decision is made here, so it is important to think well. Since I was a bit intimidated by the “I’ll configure things myself” option, and the “Standard 2 Tier Configuration” is the only one supported in the manual, I decided to go for this one first. If you are looking for a sort of tutorial, I would recommend this one, in order to check that everything is working on your system, etc.

Although they “claim” in the manual that the “client” group may correspond to many nodes, connected to one server, I found out that I could only make each client to talk to the server (and vice-verse), but I could not make the client nodes to talk to each other. It was like they were subscribing the “news” from the server, but the “news” that were arriving to the server via other nodes were not actually considered as “news”.

After that, I decided to try the “Multiple Sources to One Target Configuration”, which is also described as “Data Warehousing”. This was not exactly what I was looking for, but I was able to modify the architecture, until I arrived to something that suited me (and that I will describe later). The next screens, let you define the database connection string, and the url for communicating with the SymmetricDS instance; in my case:


(where invislaptop resolves to my server’s IP address)

After this, you are taken to the configuration dashboard, that should be “unlocked”, by using the key provided by email. The next thing you want to do, is to go to the “configuration” section. This section is very powerful, at the same time that is complicated and it allows you to tune and refine every aspect of the synchronization, with the aid of some tools for “bulk” tasks. It is certainly possible to do all this (on the FOSS version), by editing the configuration files, but I found this GUI very useful, at least for a “newbie”.

The “Data Warehousing” “pre-cooked” configuration generates a series of node groups:

  • regsrvr: registration server
  • target: target data source
  • source1: group of nodes that provide data to the target
  • source2: group of nodes that provide data to the target
  • sourceN: …

In my scenario I “left” only three nodes: the registration server, a target and a source (“source1”), and removed the other ones. The names are not so important, and I could have just called them “regsvr”, “node1” and “node2” (for instance).


The “group links” section, actually establishes the dynamics between all these groups of nodes, whatever name you called them. In my case, the registration server “waits for pulls” from both node groups (“target” and “node1”). The “target” group pushes changes to both, to the registration server and the “source1” groups. The “source1” group, pushes changes to both, the registration server and the “target” groups.


The system could be described, by something like this:


On the “routers” tab, you can define the details of these connections between nodes, through triggers (one for each action):


The triggers for each table, are defined on the “table triggers” tab.


you may defined them individually for the tables you are interested in, or do a “bulk” define by choosing “auto-create” Then, you have the option to connect the routers to the triggers on this tab, or in the “routers” tab.
When this is done, you should have a trigger for each each table, on each update/delete/remove action (according to what you have defined).

The server setup, is actually the most complex and time consuming configuration step (which I did not cover exhaustively!). After this, I went to each of my clients, and run the installation and setup again.
This time, I choose to add a “client” node instead. The “client” nodes will attempt to register during the setup, by contacting the server on the address you provide; in my case:


Unless you open the registration on the server for that particular node (by imputing its ID and group) the registration will fail. This is ok, and you can go through the entire process of creating the client, without registering the node.
When you finished the registration, if you go to the server console, and open “Manage nodes”, you will see one url under the server entry. This should be the client node, that contacted the server in order to register. If you right-click this entry, and choose “allow”, the server should be able to register the node. If you want, you may re-load the data on the client, by choosing “Send initial load to” (this actually should not be necessary, as the server should send an initial load, when allowing the node).
After registering both nodes, my setup looked like this:


After successfully registering all clients on the server, the system should be up and running. Note that you should have the symetricDS daemons running on the three nodes, to have a fully functional scenario.
I edited a record on the server, and it got replicated to the Ubuntu and Windows clients.




Then I tried to edit a record on each one of the other nodes (“target” and “client1”), and watched the changes being pushed to the other nodes. It seems that the daemon is listening for changes at very small intervals, since the changes were propagated through the system almost immediately. However I did not test it with more complex changes, including batches of data.

From this experience I would say SymmetriCDS performs quite well, and with the aid of the GUI on “symmetricDS pro”, it is not too hard to setup, once you are clear about what you are looking for and understand where to setup things. This is good because I did not find much documentation on the web apart from the simpler scenario (“Standard 2 Tier Configuration”), neither did I find posts on forums discussing this.

Furthermore it would be interesting to test this system with a “tougher” scenario: larger and more complex batches of changes, more nodes, and sometimes some (or all of) them offline. This would obviously trigger the “conflict” situation, which is also the one that “scares” me most.


Ubuntu 4 Beginners

After installing Ubuntu three times, in the past few months, and after having many requests to do it again, I have finally decided to put it all together in a workshop. It is going to be next Saturday, in Barcelona, in my favourite co-working space. And it is “free” as beer, and GNU/Linux 🙂

The “official” announcement will be tomorrow, I think, but you can be the first to read it here 😉


Ubuntu 4 Beginners

Did you ever think about installing Ubuntu, but never actually have the “courage” to do it alone? Then this workshop is for you.


In the first part I will introduce the GNU/Linux Operating System, by explaining some basic concepts and showing some applications.
The second part will focus mainly on the installation process of Ubuntu, and I will install it “live”, on a virtual machine.
At the end of the session I can help people who are interested, to perform the installation on their own computers. Note that this will be *at their own risk*!

Target Audience:
This workshop targets people with a limited knowledge of *Nix systems, although some proficiency in using computers would be nice.
If you are a proficient *nix user or developer, and are interested in specific parts of the OS (such as the kernel), you may be interested in a more advanced workshop. If you are wondering what a *nix user is, please come: this workshop is for you 🙂

If you have Ubuntu installed on your laptop, or you are planning to install it at the end of the workshop, you may bring it with you. Otherwise, laptops are not required.


Practical Info:
The duration of the workshop is approximately 2 hours (11:30h-13:30h), including a 10 minute break. Note that this is a free workshop, but you do *need to register*, in order to attend. Please do it, by filling this form: it should only take 2 minutes.
For practical reasons, I will limit the number participants to 20, on a: “first come, first served” basis.

This workshop is hosted by MOB/Made (Calle Bailen 11, Bajos. 08010 BCN) and all donations collected by the bitcoin wallet bellow will be given to Made, a non profit organization.



Quick guide to Auditing a (postgreSQL) Database: putting it all together

On my previous post, I suggested how to create a schema, a table and a trigger function, in order to audit a PostgreSQL database.
To audit a table, you would have to create a trigger for that table, calling the code from the generic trigger.
In my case, I want to audit every table in the database, and I think most people will likely want to audit every table, or at least most tables in the database.
To escape the tedious task of writing code to implement that n-times, I put together a script that will generate an audit trigger for each table in the database.If you want to apply it to a restricted number of tables instead, you could easily change it to read the table names from a list.

CREATE OR REPLACE FUNCTION create_audit_triggers()
 _string varchar ( 1000 );	

FOR r IN SELECT distinct tablename FROM pg_catalog.pg_tables where schemaname='public'  LOOP

			     FROM information_schema.triggers
			     WHERE event_object_table = r.tablename
			     AND trigger_name = r.tablename || '_audit'

				--raise info '%' , r.tablename;
				_string :=' CREATE TRIGGER ' || r.tablename || '_audit ' ||
				' AFTER INSERT OR UPDATE OR DELETE ON ' || r.tablename ||
				' FOR EACH ROW EXECUTE PROCEDURE audit.if_modified_func();';
				raise info '%', _string; 
				EXECUTE ( _string ) ; 	

	END IF ; 

end loop;

  COST 100;
ALTER FUNCTION update_info_tables2()
  OWNER TO postgres;

This will check if the trigger already exists (for which an error would be raised!), and generate the triggers during the blink of an eye (depending on the size of your database!). Thus you could use it for updating the triggers, after you added a couple of tables in the database.

Quick guide to Auditing a (postgreSQL) Database

According to Wikipedia, ‘Database auditing’ involves observing a database so as to be aware of the actions of database users. A bit like ‘spying on the user activity’.
This of course, could be useful, if you have a database with multiple users and store some sort of ‘confidential’ information.


Previously I implemented some code to do this in SQL Server, that basically involved:

  • creating a table to store the audit information.
  • create an ‘encode scheme’ for this table (i.e.: how to distinguish ‘edits’ from ‘removes’, etc).
  •  creating triggers for every table that I wanted to audit, that do the ‘house-work’ (these were of course, generated from a script)

Moreover, these ‘encoded changes’ where exported to JSON, as they were the basis for a synchronization system that I implemented,

This was working fine, until I had to port the database to another RDBMS, which gave me the opportunity to rethink the structure that I had in place.
Instead of porting directly the code from T-SQL to PSQL, I did a little research and (gladly!) found out that Postgres had an ‘inbuilt’ support for audit. It took me about 10 minutes, to ‘get grips with it”, which is what I describe next.

First thing would be to create a table to store the ‘changes’. The PostgreSQL wiki, actually advises to use a different schema, which I I agree is a good idea.

-- create a schema named "audit"
CREATE schema audit;
REVOKE CREATE ON schema audit FROM public;

CREATE TABLE audit.logged_actions (
schema_name text NOT NULL,
table_name text NOT NULL,
user_name text,
action_tstamp timestamp WITH time zone NOT NULL DEFAULT current_timestamp,
action TEXT NOT NULL CHECK (action IN ('I','D','U')),
original_data text,
new_data text,
query text
) WITH (fillfactor=100);

REVOKE ALL ON audit.logged_actions FROM public;

The stored information is: the relevant schema and table names, the username (so it is important to enforce a user policy here) and a timestamp. Then we also have the ‘type’ of action, that is one of the following: insert, delete or update. This is not supporting triggers or selects, which I guess it’s ok. Then we have the previous value and current value. For inserts, we have a change from nothing to something; for deletes we have a change from something to nothing and from updates we have a change of something to something. In a way, you could figure out the change type from looking at these values, so the ‘action’ field is perhaps a bit redundant; but then, you would have to represent ‘nothing’ as a sort of special value (or keyword) and not an empty space (that could be misunderstood by an empty string).
Finally we have the ‘query’, which is the exact query that triggered this audit. Although this is also a bit redundant, since it could be reconstructed from the other values, it is not a bad idea since it allows to quickly see/reproduce exactly what happened.

Then we can create some indexes:

CREATE INDEX logged_actions_schema_table_idx
ON audit.logged_actions(((schema_name||'.'||table_name)::TEXT));

CREATE INDEX logged_actions_action_tstamp_idx
ON audit.logged_actions(action_tstamp);

CREATE INDEX logged_actions_action_idx
ON audit.logged_actions(action);

The next step is to create the trigger to ‘fill’ this table, on relevant actions:

v_old_data TEXT;
v_new_data TEXT;
/*  If this actually for real auditing (where you need to log EVERY action),
then you would need to use something like dblink or plperl that could log outside the transaction,
regardless of whether the transaction committed or rolled back.

/* This dance with casting the NEW and OLD values to a ROW is not necessary in pg 9.0+ */

v_old_data := ROW(OLD.*);
v_new_data := ROW(NEW.*);
INSERT INTO audit.logged_actions (schema_name,table_name,user_name,action,original_data,new_data,query)
VALUES (TG_TABLE_SCHEMA::TEXT,TG_TABLE_NAME::TEXT,session_user::TEXT,substring(TG_OP,1,1),v_old_data,v_new_data, current_query());
v_old_data := ROW(OLD.*);
INSERT INTO audit.logged_actions (schema_name,table_name,user_name,action,original_data,query)
VALUES (TG_TABLE_SCHEMA::TEXT,TG_TABLE_NAME::TEXT,session_user::TEXT,substring(TG_OP,1,1),v_old_data, current_query());
v_new_data := ROW(NEW.*);
INSERT INTO audit.logged_actions (schema_name,table_name,user_name,action,new_data,query)
VALUES (TG_TABLE_SCHEMA::TEXT,TG_TABLE_NAME::TEXT,session_user::TEXT,substring(TG_OP,1,1),v_new_data, current_query());
RAISE WARNING '[AUDIT.IF_MODIFIED_FUNC] - Other action occurred: %, at %',TG_OP,now();

WHEN data_exception THEN
WHEN unique_violation THEN
LANGUAGE plpgsql
SET search_path = pg_catalog, audit;

The ‘TG_OP’ variable inside a trigger, is a string that tells us for which operation the trigger was fired (INSERT, UPDATE, or DELETE). The rest is fiddling around with the ‘old’ and ‘new’ values.

Actually this is all to it, regarding having the audit ‘structure’ in place for postgresql. To put it ‘in action’, auditing a table is as simple as this:

CREATE TRIGGER fr_frame_audit
FOR EACH ROW EXECUTE PROCEDURE audit.if_modified_func();

In the example above, I created a trigger for auditing table ‘fr_frame’ (of course you can create a sp procedure to generate these statements, if you want to generate a trigger for auditing every table in the database…).
Then I went to table ‘fr_frame’ and deleted a row. It got stored like this:

"public";"fr_frame";"postgres";"2014-02-11 12:36:52.063113+00";"D";"(67,"bin frame",missing,missing,1,1,missing)";"";"DELETE FROM public.fr_frame WHERE id = '67'::integer"

Then I modified and added a row; it got stored like this:

"public";"fr_frame";"postgres";"2014-02-11 12:38:54.793902+00";"U";"(59,"Sampling Frame",missing,"Initial Sampling frame",1,2,missing)";"(59,"Sampling Frame","Sampling Frame","Initial Sampling frame",1,2,missing)";"UPDATE public.fr_frame SET nameeng='Sampling Frame'::text WHERE id = '59'::integer"
"public";"fr_frame";"postgres";"2014-02-11 12:39:14.290898+00";"I";"";"(68,test,test,test,1,2,test)";"INSERT INTO public.fr_frame(name, nameeng, description, id_cloned_previous_frame, id_source, comments) VALUES ('test'::text, 'test'::text, 'test'::text, '1'::integer, '2'::smallint, 'test'::text)"

As I said before, the encode of non-existing values (nothing) as an empty string is not the most accurate approach, but since we have more information to complement it, it works. Also, the fact that it is row-based rather than field-based (as I had in my implementation), originates the serializing arrays, which is not exactly normalized… on the other hand, I can accept it from the point of view of storage.

Overall I think I accept it as a good solution, measuring all the pros and cons, and high fives for being so simple to implement.

You can find this information (and more), in the PostgreSQL wiki.

Building Cross-plattform Applications (for real!)

I have been writing many posts about the Qt library, without making a proper introduction.
Qt is a cross-plattform C++ framework, that provides support for a lot of things such as UI, multi-threading, Graphics, etc. Since C++ itself does not have so many libraries built in as you would find on other libraries, it is a good idea to use something like this. There are other frameworks specific things such as graphics (for instance GTK+), but I don’t know of any as complete as Qt. The .NET framework is quite complete, but unlike Qt it is not cross-platform, neither it has a permissive GPL/LGPL license, so if you care about these two things it is clearly *not* an option.

Having said this, I am quite happy with the Qt libraries in the long run, even things are not as easy as they would seem.

The Windows and Linux environment are different enough, even if you stick to C++ and Qt.
In my Linux environment, I use g++ make (the default compiler), and let the system (aka package manager) to decide what is the most appropriated version. It actually takes care of all the environmental variables and I do not have to worry too much about setting a build environment, whether I use Qt creator (the “native” Qt IDE) or just the command line.
On the other hand, Windows has got a couple of options regarding compilers:

  • there is the mingGW compiler, which people use for portability (but I don’t particularly like it since you need to setup a whole set of tools, that are not native on Windows)
  • there is the native Microsoft Visual Studio toolchain, which is by excellence, a “Windows compiler”
  • there is the Qt creator “jom” compiler, which allows using multi core, but somehow I am a bit reluctant in using it, because I don’t see anybody using it outside Qt creator.

The thoughts and “suspicions” above, are nothing else but that: thoughts and suspicions. Since I have been programming in Windows for quite a while using Microsoft Visual Studio and I am quite happy about it, I just decided to use it in my Windows projects.

If you want to stick to basic configurations, the differences between Windows and Linux projects may be small, but as soon as you start to complicate them a bit it is not so simple anymore. Recently I decided to link my application statically, in order to ease deployment in a complete “user-proof” scenario. These required rebuilding Qt statically, in both OS.
After successfully compiling Qt, I tried to build my project on Linux using Qt creator and it “worked as a charm”. I did not have to change any of the environment variables, but only have to point to which version of Qt I wanted to use inside Qt creator. There were only a few things that I had to keep in mind:

  • My application uses a plugin, so I had to also link it statically (creating a .la)
  • There were some minor changes in the qmake project files, *both* of the plugin and the application, in order to compile them statically.
  • It is better to clean/rebuild the projects, so that we don’t run into the risk of having files *left* from a previous dynamic linking. From my experience, “make clean” is not so tidy, so I would recommend going inside the directories and remove the .obj, or any other intermediate files by hand…

This is the only change that I had to make in the project file of my designer plugin

dynamic linking:

CONFIG += debug_and_release

static linking:

CONFIG += release staticlib

(with static linking, there is generally no point on debugging, so I am generating only a release build)

Then I could compile my application, and link against the static versions of Qt and of this plugin, by slightly modifying the project file:

dynamic linking:

CONFIG += debug_and_release

static linking:

CONFIG += release static

And that was about it: I got a binary file with 26.9 MB, that I can take with me to any Ubuntu system 🙂

Now the Windows part was a bit more painful. First I could not get the Qt Creator to work, since it did not correctly pick up the VC+ settings, and complained about the static version of Qt not having been built with the same compiler I was trying to use. To speed up things, I decided to compile the application on the command line, more specifically inside the Visual studio shell, which is the same place where I compiled Qt.

I had some persistent linking errors regarding a DLL linkage with my plugin. I read somewhere that by default in Windows, qmake will attempt a dll linkage unless its explictly told otherwise; thus I added this flag to the DEFINES of my project files:


In Linux I did not run into such a problem. In any case, it did not solve the linking errors. After researching a bit more, I found out that to call a plugin statically, you have to invoke a specific macro on the “main” of the application, *and* include QtPlugin. Another thing that was not necessary in Linux…

#include <QtPlugin>;


And, finally the plugin has to be *explicitly* added to the project file, in this way:

QTPLUGIN += catchinputctrl

I still had linking errors, this time regarding *not* finding the plugin library; the directive that I had in Linux did not worked, and I messed around a bit with the “-L” and “-l” options in LIBS but without success, so I ended up copying the .lib file to my project directory (a quick fix). After that, I could generate the binary, but not without some linking warnings:

Creating library ..\release\faocas.lib and object ..\release\faocas.exp
frmcatch.obj : warning LNK4217: locally defined symbol ??0CatchInputCtrl@@QAE@PA
VQWidget@@@Z (public: __thiscall CatchInputCtrl::CatchInputCtrl(class QWidget *)
) imported in function "public: void __thiscall Ui_FrmCatch::setupUi(class QWidg
et *)" (?setupUi@Ui_FrmCatch@@QAEXPAVQWidget@@@Z)
frmoperation.obj : warning LNK4049: locally defined symbol ??0CatchInputCtrl@@QA
E@PAVQWidget@@@Z (public: __thiscall CatchInputCtrl::CatchInputCtrl(class QWidge
t *)) imported
frmtrip.obj : warning LNK4049: locally defined symbol ??0CatchInputCtrl@@QAE@PAV
QWidget@@@Z (public: __thiscall CatchInputCtrl::CatchInputCtrl(class QWidget *))
frmcatch.obj : warning LNK4217: locally defined symbol ??1CatchInputCtrl@@UAE@XZ
(public: virtual __thiscall CatchInputCtrl::~CatchInputCtrl(void)) imported in
function "public: virtual void * __thiscall CatchInputCtrl::`scalar deleting des
tructor'(unsigned int)" (??_GCatchInputCtrl@@UAEPAXI@Z)
frmoperation.obj : warning LNK4049: locally defined symbol ??1CatchInputCtrl@@UA
E@XZ (public: virtual __thiscall CatchInputCtrl::~CatchInputCtrl(void)) imported

frmtrip.obj : warning LNK4049: locally defined symbol ??1CatchInputCtrl@@UAE@XZ
(public: virtual __thiscall CatchInputCtrl::~CatchInputCtrl(void)) imported
mt.exe -nologo -manifest "release\faocas.intermediate.manifest" -outputr

The output dir contained my executable, as well as a exports library file and a copy of the lib. These are unnecessary, and I can happily copy my 12.2MB binary around, without having to ship anything else.

Some questions that remain in my mind:

  • Why are the binaries generated by Windows and Linux so different in size? (one almost *doubles* the size of the other)
  • Why is the static and dynamic linking in Windows so different?
  • Why is the static linking in Windows so different from the one in Linux, and why is this so undocumented? (does anybody use it at all??)
Static linked app in Windows

Static linked app in Windows

Static linked app in Linux

Static linked app in Linux

Static Linking?

From time to time, I have this moments when I cannot deploy my application properly and decide that I want to link it statically (then I generally give up, because it requires me to link the Qt libraries statically…). But is it really better to prefer static over dynamic linking?

As in so many other cases, it depends on what you want to do. I read that in terms of performance, there are trade-offs in both approaches, so in the end it really does not matter so much. From my point of view, the biggest advantage of static linking is the fact that you can ship one single file with your application, removing the risk of “broken” dependencies. That is, in terms of deployment, quite an advantage!

On the other hand, if everybody would link statically, we would literally have “thousands” of libraries “repeated” inside our system, packed inside “huge” binaries. It does not make much sense, does it?

Dynamic libraries are also “cool”, because we can (till a certain extent) replace them by newer (improved) versions, without having to recompile our application. That is like a huge benefit, in terms of “bug fixing” of third party libraries.

After removing the performance issue, my verdict would be:

  • For myself, I would like to minimize resource consumption by using as much as possible, shared libraries (dynamic linking).
  • For “bullet proof” systems, where users are not experienced in installing software, and are likely to “mess up” the system by removing parts of it, I would consider providing them statically compiled versions of the software, instead. The software will likely be “bigger “(although there are tools to minimize this, such as UPX), and a bit more “hungry” of resources, but this is also the only way to prevent the DLL hell.

Finally, it is important to mention that the type of linking may be conditioned by licensing issues.  For instance due to the “nature” of the license, GPL libraries would “contaminate” any software statically linked with them.

Cross-compiling, using Qt

On a previous post I described a simple example, of how-to cross compile an Windows application in a Linux environment.
Although this worked fine and raised my motivation, it is not a very useful example since the real world is much more complex. One of the complexities that I want to introduce is the use of Qt libraries, that are used thoroughly in my projects. I will now go in detail, through a slightly more complicated “Hello World”, where I make use of these libraries.

A key thing to understand, when cross compiling with Qt, is that you need to setup the development environment, both in the host and in the target OS (those are in this case: Ubuntu and Windows).

I started by downloading the Qt distribution, compiled for MinGW (of course, you may also decide to download the source codes and compile it yourself, which will add another complexity layer to this task…):

wget http://download.qt-project.org/official_releases/qt/4.8/4.8.5/qt-win-opensource-4.8.5-mingw.exe

Then I ran this with Wine, and followed the installation instructions:

wine qt-win-opensource-4.8.5-mingw.exe

Qt got installed under my “wine drive”, in the $HOME directory:


As I mentioned before, you also need a working Qt environment for Linux. I had Qt 4.8.7 installed in my machine, and I stupidly thought I could use that; it turns out I can not. The Qt versions in Windows and Linux need to match! Since MinGW did not released a 4.8.7 version yet, I decided to download the Linux version of Qt 4.8.5, and installed it along with my current version (let us see what problems this may bring me in the future, if I keep developing in 4.8.7!).
If you did not understand my grump in the last paragraph, the only thing you need to know is that I downloaded Qt 4.8.5 for Linux, and installed it in:


The next step is to configure the mkspecs for cross-compiling; for that you need to navigate to /usr/local/Trolltech/Qt-4.8.5/mkspecs and create a new directory there. I called it “win32-x-g++”.


Inside this directory, you need to place a version of qmake.conf, with all the necessary information for cross compiling; mainly you need to pay attention to the paths of the Qt installed in Wine, and the minGW paths in Linux. Here is a working version for me; please amend it to reflect your paths.

Then I wrote a very simple Qt “Hello World”:

// Hello World in C++ for the Qt framework
int main(int argc, char *argv[])
	QApplication a(argc, argv); 
	QLabel l("Hello World!", 0); 
	l.resize(300, 200); 

This is just a Window with “Hello World” written, but it is calling the Qt Libraries.
My Qt project file, looks like this (I generated it with qmake, and then added the relevant libraries to the Qt variable):

# Automatically generated by qmake (2.01a) Fri Dec 20 13:18:06 2013

TARGET = hello
QT += core gui
CONFIG += qt

# Input
SOURCES += hello.cpp

I make a small parenthesis here, to note that the qmake version should be the one on:


If you do have more than one Qt version (like I do!), it is worth to set the $QTDIR variable to this directory, while cross compiling. You can check if this working as expected, by checking the path of qmake:

which qmake

If this points to /usr/local/Trolltech/Qt-4.8.5, then everything is ok; otherwise, please take a moment to export your $QTDIR.

To generate the makefiles, using the created mkspecs file above, you can use the following syntax:

qmake -spec win32-x-g++

Then you can compile the application, using make. If the compilation was successful, you should be able to run the executable “hello.exe”, using wine. When I attempted to do this, I had some errors that related to not finding some libraries. As in my previous example, I copied “mingwm10.dll” to the current directory, but I still had errors related to Wine not finding the Qt libraries.

I edited the QTDIR and PATH variables on Wine(Windows), by calling regedit.


Unfortunately that still did not do it for me. I thought a quick fix, for forcing the application to pick the libraries would be to copy the required DLL’s to the app path. Unfortunately that also did not worked 😦

err:module:import_dll Library libgcc_s_dw2-1.dll (which is needed by L"Z:\\home\\joana\\projects\\cross-compiling\\hello\\release\\QtCore4.dll") not found
err:module:import_dll Library QtCore4.dll (which is needed by L"Z:\\home\\joana\\projects\\cross-compiling\\hello\\release\\hello.exe") not found
err:module:import_dll Library libgcc_s_dw2-1.dll (which is needed by L"Z:\\home\\joana\\projects\\cross-compiling\\hello\\release\\QtCore4.dll") not found
err:module:import_dll Library QtCore4.dll (which is needed by L"Z:\\home\\joana\\projects\\cross-compiling\\hello\\release\\QtGui4.dll") not found
err:module:import_dll Library libgcc_s_dw2-1.dll (which is needed by L"Z:\\home\\joana\\projects\\cross-compiling\\hello\\release\\QtGui4.dll") not found
err:module:import_dll Library QtGui4.dll (which is needed by L"Z:\\home\\joana\\projects\\cross-compiling\\hello\\release\\hello.exe") not found
err:module:LdrInitializeThunk Main exe initialization for L"Z:\\home\\joana\\projects\\cross-compiling\\hello\\release\\hello.exe" failed, status c0000135

Then I thought, that even if I cannot run the executable with Wine, it does not necessarily mean that I did not generate it correctly (which is ultimately my purpose, with all of this!). So I went to a Windows guest in VirtualBox, and I managed to successfully run the application! 🙂


This was a nice outcome, however it is still necessary to do this for the entire project, tracking all required Qt dll’s, Boost, and any other needed libraries. It would probably be a good idea to consider static linking, since I would not have to ship the DLL’s, and would also avoid digging into the problem of “why Wine is not finding the shared libraries”. However, from my experience, static linking of the Qt framework is quite an onerous task, and in this case I would have to do it twice (for Linux and for Windows).

“Hello World” Cross-Compiling

I recently switched to work on a Linux environment, in a project where most of the potential users will be using Windows.
Theoretically that is not a problem, since I am sticking to standard C++, and a few other libraries that are fully cross-platform (e.g.: Qt[1], Boost[1])
Of course in practice, things are always slightly more complicated 🙂

Since my original project was created in Windows, using the nmake compiler, I still have a functional environment that I can use. However this solution requires me to:

  1. start Windows;
  2. checkout the latest source code;
  3. tweak the configuration for the Windows environment;

It is mainly 1. and 3. that bother me; I don’t start Windows that often, and there are quite a few things to tweak, when you change from GNU g++ to MS nmake[3].

There are many options to make things a bit “smoother”, and one of them would be to change the compiler to Jom[4] or MinGW, so that it could be use in both OS, with minimal tweaks. Another option would be to create a repository only for the source code, and leave the configuration files unchanged, both in Windows in Linux. These are all things that I want to explore, but for the moment I ceased to my “fascination” with cross-compiling, and decided to investigate it a bit further. Wouldn’t it be so cool to be able to generate a Windows binary, straight from Linux? It seems like the perfect option, if it was that easy!

To get a functional cross-compiling environment, you basically will need Wine[5] and MinGW[6]. In a nutshell, Wine is a is a compatibility layer capable of running Windows applications on several POSIX-compliant operating systems; the acronym says everything: “Wine Is Not an Emulator”. MinGW is a minimalist development environment for native Microsoft Windows applications. Installing these software in Ubuntu, is as easy as:

sudo apt-get install wine mingw32 mingw32-binutils mingw32-runtime

This should be it; so I started with a simple example that I grabbed from [7]:


int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance,
    LPSTR lpCmdLine, int nCmdShow)
    "Cette fenêtre prouve que le cross-compilateur est fonctionnel !",
    "Hello World", MB_OK);
  return 0;

I compiled it like this, using MingGW:

i586-mingw32msvc-g++ -o essai.exe essai.cpp

To execute the binary, you will need the mingw library: mingwm10.dll, that can be copied like this to the current directory:

gunzip -c /usr/share/doc/mingw32-runtime/mingwm10.dll.gz &gt; mingwm10.dll

Then the application can be launched with:

wine essai.execute


[1] http://qt-project.org/
[2] http://www.boost.org/
[3] http://msdn.microsoft.com/en-us/library/dd9y37ha.aspx
[4] http://qt-project.org/wiki/jom
[5] http://www.winehq.org/
[6] http://mingw.org/
[7] http://retroshare.sourceforge.net/wiki/index.php/Ubuntu_cross_compilation_for_Windows

“Obscure” Databases in Detail

Carrying on a bit more detailed analysis from my previous post, I will try to point some flaws that I saw in databases stored within a Relational Database Management System (RDBMS), and discuss some solutions.

To start with the basics, why using a relational model at all to represent the data?

Skipping the obvious answer that the database is stored already using a relational engine, “normalization” (Codd, 1970) is the adoption of “normal forms of relations” with the goal of avoiding certain data anomalies that can occur in unnormalized tables. In other words, being very strict with the way of representing data is going to protect the consistency and quality of the dataset we are building. On the other hand, not being strict with the data model (or not enforcing a data model at all) may create a scenario with poor quality data, where we cannot decode the information we have, which technically means it is “lost” (even if it is physically stored in the database). Think about this as an unbreakable safe box, for which you have lost the only existing key in the entire world.

The normal forms are an attempt to make sure that you do not destroy true data or create false data in your database, so it is always a good idea to go through them when designing (or redesigning) a schema.

1. Representation of arrays

One of the “bottlenecks” in the relational model, specially for people that come from a programming background is the representation of lists or groups. Sometimes people fake the representation of arrays, by repeating a group of columns, or “flattening out” the array. Apart from the data storage overhead, this is not conform with the NF1 principle. Let us look at an example.

Table “Students”

dprt student
Geography John
Geography Martin
Geography Maria

In this example, we cannot remove a course (or update it) without involving a series of rows.

Sometimes arrays are also represented by encoding a list inside a text field, for instance separated by commas. This is also not conform with normalization, since this data type is non-existent and it is breaking the principle that the combination of row, column in a table should always refer to one record. The result is a lot of string-handling procedures to work around this kludge.

Table “Students”

dprt students
Geography John, Martin, Maria

In this example, to delete (or update) a student name we need to parse and modify the students string which is so inconvenient that I don’t need to explain further why it is not a good idea to have this approach.

The proper way to represent arrays in this case is to split the students info in two tables, and relate them with a key.

Table “Departments”

id name
1 Geography

Table “Students”

id id_dprt name
1 1 John
2 1 Martin
3 1 Maria

The two tables above are in the normalized form, and we can operate over them using the relational model. Generally, when things involve complicated procedures to work, it means they are prone to errors, so we should keep everything as simple as possible. With the table structure above, I can implement a cascade delete that will remove all the students in the Geography Department, once I close this department. This goes the same as saying, that once I close the Department, there will be no students enroled in this department lying around the database (I call them “ghost” records).

2. Use of Natural Keys

Natural keys in themselves are not discouraged. They have a meaning to the user, which is generally a good idea, and their use prevents generating extra columns on purpose to be keys. Automatic keys (like identity fields) are generally not conform with the normalized model since they bring some dependence from the physical model (and may be implemented differently by different vendors) and they bring replication issues. With the computer power that we have nowadays it is also not a such problem to have a key involving multiple columns.

However these natural keys may be “hiding” other problems, that are themselves a “threat” to a normalized database.

2.1 Are these keys “preventing” implementing the relation between tables?

In my University database, I can represent the entities: universities, departments, courses and students as different tables. Since I know, there are one-to-many relationships in Universities->Departments, Departments->Courses and Courses->Students, I can enforce this idea by using foreign keys constraints.

CREATE TABLE Universities (
    id        integer,
    name       varchar(40),

CREATE TABLE Departments (
    id        integer,
    id_universities integer references Universities(id),
    name       varchar(40),

    id        integer,
    id_dprt integer references Departments(id),
    name       varchar(40),
    CONSTRAINT pk_courses PRIMARY KEY(id)

    id        integer,
    id_courses integer references Courses(id),
    name       varchar(40),
    CONSTRAINT pk_students PRIMARY KEY(id)

If I want to know the University of a student named ‘Martin’, I just have to build my query knowing the “path” from Student to University.

Select Universities.name, Students.name
from Universities inner join Departments ON Universities.id=Departments.id_university
inner join Courses ON Departments.id=Courses.id_dprt
inner join Students ON Courses.id=Students.id_course
WHERE student.name like 'Martin'

(DISCLAIMER: please forgive me if there is some error in this query: I did not actually test it, but I think this is sufficient to get the general idea…)

On the other hand, if I don’t implement this structure in the database, and I still want to know the University of a named Student, or the University of a named course, I can do something like this.

CREATE TABLE Universities (
    name       varchar(40),
    CONSTRAINT pk_univ PRIMARY KEY(name)

CREATE TABLE Departments (
    university_name varchar(40),
    name       varchar(40),
    CONSTRAINT pk_dprt PRIMARY KEY(name)

    university_name varchar(40),
    name       varchar(40),
    CONSTRAINT pk_course PRIMARY KEY(name)

    university_name varchar(40),
    name       varchar(40),
    CONSTRAINT pk_students PRIMARY KEY(name)

In this case, I can know the University of any student (or any course), by querying a single table. The “natural key”: university_name, will provide the answer to my question. This design may run into problems, for instance if I have two students with the same name, and on the same University, but enrolled on different departments. That problem could be “solved” by bloating the Students table with a two-column key:

    university_name varchar(40),
    course_name varchar(40),
    name       varchar(40),
    CONSTRAINT pk_students PRIMARY KEY(university_name,course_name)

It is easy to imagine other violations of uniqueness or other requests of information, that could lead us to add more columns to the primary key. My whole point here is that this is a strategy to avoid implementing a relational structure in the database, and therefore it is conceptually wrong. If we know the relations between entities, we should implement them exactly as they are, and not approach the database as a set of self contained-tables.

Actually this behaviour can be tracked to people who first worked with file systems, using tapes. They tend to design one huge file and do all the work against these records; this made sense at the time, since there was no reasonable way of joining a number of small files together without having the computer mount and dismount lots of tapes. With RDBMS there is absolutely no reason to avoid joins between tables. A normalized database tends to have a lot of tables with a small number of columns per table.
If we cannot produce this sort of output because we don’t know the relations between the entities we are modelling, than I would actually question the use of a database at all (or at least, a relational database).

2.2 Do we have information represented in more than one place?

In the example above, we have the natural key “university_name” repeated all over the database. Note that there are no foreign keys, and therefore there is no guarantee that the university names on tables “Courses” and “Students” match the names on table “University”. It is easy to imagine why we may run into problems with this approach. The key that determines one attribute should be in only one table and therefore the attribute should be with it. All the other tables that “mention” this attribute should refer to this table.
Apart from the fact that we are not “repeating” information all over the database and “bloating” the storage, this is the only way of guaranteeing consistency among the dataset. If a fact appears more than once, one of the instances is likely to be in error.

The database is only good in “safeguarding” things that we implement. Since the table “Students” is not formally connected to table Courses, nothing prevents me from having a course listed in students, that is not listed at all in table “Courses”. This may be, either because I did not introduce it there, or because I misspelled it (or some other reason that I can’t remember right now). If we use an application to interface with the database, we could implement strategies to prevent this kind of errors, but actually:
why should we put this responsibility in the “shoulders” of the front-end when it could be enforced by design, in the first place?

3. Table and Column Names

Although there are not strict conventions for naming tables and columns, we should stick to the idea that tables represent either “entities” or “relationships” and the columns represent attributes, and named them accordingly. Since they store sets of rows, it is expressive if the name is in plural (like “Students” or Universities”). Actually this may seem like being over-careful, but there are not many things as bad in terms of data loss, as having a database where we don’t know what the tables or the attributes represent (and not having any enforced constraints only makes it worst).
It is also a good idea to keep the names consistent across the namespace. For instance if we have a column named “temperature” and another one “called “watertemp” and they both refer to the same thing (temperature of the water), it is generally a good idea to use the same name. This, of course, goes against the paradigm of a database as a repository of data from different sources where we don’t even change the field names when importing. But with that kind of database, there is very little you can do without a tremendous amount of work (and possible even then…).
Regarding naming conventions, everything that makes the database easier to understand and to work with, should be implemented.

4. Use of Nulls

The use of nulls brings a series of issues and should be avoided as much as possible, for instance by providing a default value or a meaningful code (“missing”, “invalid”, etc). Having many nulls on a table, may be a sign that it has not been “normalized” properly (for instance the “flattening” of an array). Nulls should *only* be used for a value that it is missing now, but may be resolved later.

5. Information that should not be stored in the database

My final thoughts go to the “temptation” of mixing the contents of the database with the mechanisms to manage this information (for instance the front-end that we use to visualize it and populate it). A database models a certain reality (for instance a University, or an Address Book). Introducing “auxiliary” tables that are tied to the front-end is possible, but not without “polluting” the dataset with information about how we deal with the dataset, which is not really part of the dataset itself. Frontends may change, and ways of visualizing data may change, but our data layer should be kept conveniently “abstracted” from all these things.
This is a reflexion that I make, that may justify the use of n-databases instead of one, where we can relate them to each other. The extra complexity required by the connection mechanism can be justified by the clarity of the system, and the simplicity of dealing only with the parts that we need.

Welcome to the “Obscure” World of Databases

How would we picture the image of a non doctor, entering an operations room and performing a surgery? Probably pretty badly. After removing the scope of the consequences, to me this is very similar to the image of a non database person, designing a database.

For several reasons, people who have not been formally trained as computer scientists find themselves performing IT tasks, some of them very sophisticated. I don’t see anything wrong with people that were not formally trained embrassing an IT career (actually I am one of them), as long as… they do embrace it. That is: study or research the things they don’t know, and do care about standards of quality. If they don’t do it, then they are just bringing into “shame” the name of everyone who is working on the field, by bringing the level really down.

All these thoughts – that are actually recurrent in my life – were triggered by analysing what I thought it was a relational database. By going through it and refactoring it, I got a pretty good collection of the things that you should not do, when designing a database. These are some ideas that I would like to clarify; if you think that they are obvious, you would be surprised with what I saw.

Choosing a relational database management system (RDBMS), does not automatically mean that you have a relational database. A relational database is a database that is organized in terms of the relational model, as formulated by Edgar F. Codd. According to Wikipedia[1], the purpose of this model is to “provide a declarative method for specifying data and queries: users directly state what information the database contains and what information they want from it, and let the database management system software take care of describing data structures for storing the data and retrieval procedures for answering queries”. It is a requirement of the model that the users state exactly what the database contains, and describe properly how it is organized; only in this way, it is possible to query the database and obtain exact answers to our queries.

If the database is a repository of information, some of it unkown or unnecessary, and not really organized, that means that we are not modelling the data, but only storing information. The outcome of this, is that we cannot query the database and produce any usefull answers about our data (there are no miracles!). We can store the information, but not in a much better way than if we were storing it in a filesystem. Users of this repository may implement themselves ways of dealing with this data by creating relations on-the-fly, either by using the SQL engine, or by pullling the data out and do it somewhere else. However they are not gaining anything from the data model, as they have to figure out themselves how the data is organized, and there is no guarantee that any two people would not do it in a different way.

My first bullet point is that there is absolutely no point in using a relational database engine and *not* implement a relational model; it is probably a worst solution than using a filesystem, because it may pass the wrong idea to people: that there is a relational model.

If people are not implementing the relational model because they do not know the data, or because the data is not organized in a relational manner, this is an explanation I can accept. In fact I think for many cases (and some of them which are being approached relationally) the rigid structures of a top-down design are not the providing a good solution, because knowing everything (or at least a lot) about the data is a weak assumption. For this cases there is NoSQL[2], and specially the document-driven databases provide a very flexible model, close to what people implement with filesystems. In other words: this is ok, but then don’t use a relational database engine.

On the other hand, if people are not implementing the relational model because they don’t know it, then they should learn about it. They should learn about it, at least to be able to decide that they don’t use it and go for a NoSQL database. Finally if they do not want to learn about databases, or relational models, because they are “just” biologists or economists, that is also fine but then please don’t let them design a database, specially if it is one where you want to store valuable data. And I am afraid of the institutions who let this sort of thing happen.



[1] http://en.wikipedia.org/wiki/Relational_database

[2] http://en.wikipedia.org/wiki/NoSQL