Should we overload the meaning of configuration settings? - configuration

Imagine an instance of some lookup of configuration settings called "configuration", used like this:
if(! string.IsNullOrEmpty(configuration["MySetting"])
{
DoSomethingWithTheValue(configuration["MySetting"]);
}
The meaning of the setting is overloaded. It means both "turn this feature on or off" and "here is a specific value to do something with". These can be decomposed into two settings:
if(configuration["UseMySetting"])
{
DoSomethingWithTheValue(configuration["MySetting"]);
}
The second approach seems to make configuration more complicated, but slightly easier to parse, and it separate out the two sorts of behaviour. The first seems much simpler at first but it's not clear what we choose as the default "turn this off" setting. "" might actually a valid value for MySetting.
Is there a general best practice rule for this?

I find the question to be slightly confusing, because it talks about (1) parsing, and (2) using configuration settings, but the code samples are for only the latter. That confusion means that my answer might be irrelevant to what you intended to ask. Anyway...
I suggest an approach that is illustrated by the following pseudo-code API (comments follow afterwards):
class Configuration
{
void parse(String fileName);
boolean exists(String name);
String lookupString(String name);
String lookupString(String name, String defaultValue);
int lookupInt(String name);
int lookupInt(String name, int defaultValue);
float lookupFloat(String name);
float lookupFloat(String name, float defaultValue);
boolean lookupBoolean(String name);
boolean lookupBoolean(String name, boolean defaultValue);
... // more pairs of lookup<Type>() operations for other types
}
The parse() operation parses a configuration file and stores the parsed data in a convenient format, for example, in a map or hash-table. (If you want, parse() can delegate the parsing to a third-party library, for example, a parser for XML, Java Properties, JSON, .ini files or whatever.)
After parsing is complete, your application can invoke the other operations to retrieve/use the configuration settings.
A lookup<Type>() operation retrieves the value of the specified name and parses it into the specified type (and throws an exception if the parsing fails). There are two overloadings for each lookup<Type>() operation. The version with one parameter throws an exception if the specified variable does not exist. The version with an extra parameter (denoting a default value) returns that default value if the specified variable does not exist.
The exists() operation can be used to test whether a specified name exists in the configuration file.
The above pseudo-code API offers two benefits. First, it provides type-safe access to configuration data (which wasn't a stated requirement in your question, but I think it is important anyway). Second, it enables you to distinguish between "variable is not defined in configuration" and "variable is defined but its value happens to be an empty string".
If you have already committed yourself to a particular configuration syntax, then just implement the above Configuration class as a thin wrapper around a parser for the existing configuration syntax. If you haven't already chosen a configuration syntax and if your project is in C++ or Java, then you might want to look at my Config4* library, which provides a ready-to-use implementation of the above pseudo-code class (with a few extra bells and whistles).

Related

What is the best way to handle versioning using JSON protocol?

I am normally writing all parts of the code in C# and when writing protocols that are serialized I use FastSerializer that serializes/deserializes the classes fast and efficient. It is also very easy to use, and fairly straight-forward to do "versioning", ie to handle different versions of the serialization. The thing I normally use, looks like this:
public override void DeserializeOwnedData(SerializationReader reader, object context)
{
base.DeserializeOwnedData(reader, context);
byte serializeVersion = reader.ReadByte(); // used to keep what version we are using
this.CustomerNumber = reader.ReadString();
this.HomeAddress = reader.ReadString();
this.ZipCode = reader.ReadString();
this.HomeCity = reader.ReadString();
if (serializeVersion > 0)
this.HomeAddressObj = reader.ReadUInt32();
if (serializeVersion > 1)
this.County = reader.ReadString();
if (serializeVersion > 2)
this.Muni = reader.ReadString();
if (serializeVersion > 3)
this._AvailableCustomers = reader.ReadList<uint>();
}
and
public override void SerializeOwnedData(SerializationWriter writer, object context)
{
base.SerializeOwnedData(writer, context);
byte serializeVersion = 4;
writer.Write(serializeVersion);
writer.Write(CustomerNumber);
writer.Write(PopulationRegistryNumber);
writer.Write(HomeAddress);
writer.Write(ZipCode);
writer.Write(HomeCity);
if (CustomerCards == null)
CustomerCards = new List<uint>();
writer.Write(CustomerCards);
writer.Write(HomeAddressObj);
writer.Write(County);
// v 2
writer.Write(Muni);
// v 4
if (_AvailableCustomers == null)
_AvailableCustomers = new List<uint>();
writer.Write(_AvailableCustomers);
}
So its easy to add new things, or change the serialization completely if one chooses to.
However, I now want to use JSON for reasons not relevant right here =) I am currently using DataContractJsonSerializer and I am now looking for a way to have the same flexibility I have using the FastSerializer above.
So the question is; what is the best way to create a JSON protocol/serialization and to be able to detail the serialization as above, so that I do not break the serialization just because another machine hasn't yet updated their version?
The key to versioning JSON is to always add new properties, and never remove or rename existing properties. This is similar to how protocol buffers handle versioning.
For example, if you started with the following JSON:
{
"version": "1.0",
"foo": true
}
And you want to rename the "foo" property to "bar", don't just rename it. Instead, add a new property:
{
"version": "1.1",
"foo": true,
"bar": true
}
Since you are never removing properties, clients based on older versions will continue to work. The downside of this method is that the JSON can get bloated over time, and you have to continue maintaining old properties.
It is also important to clearly define your "edge" cases to your clients. Suppose you have an array property called "fooList". The "fooList" property could take on the following possible values: does not exist/undefined (the property is not physically present in the JSON object, or it exists and is set to "undefined"), null, empty list or a list with one or more values. It is important that clients understand how to behave, especially in the undefined/null/empty cases.
I would also recommend reading up on how semantic versioning works. If you introduce a semantic versioning scheme to your version numbers, then backwards compatible changes can be made on a minor version boundary, while breaking changes can be made on a major version boundary (both clients and servers would have to agree on the same major version). While this isn't a property of the JSON itself, this is useful for communicating the types of changes a client should expect when the version changes.
Google's java based gson library has an excellent versioning support for json. It could prove a very handy if you are thinking going java way.
There is nice and easy tutorial here.
It doesn't matter what serializing protocol you use, the techniques to version APIs are generally the same.
Generally you need:
a way for the consumer to communicate to the producer the API version it accepts (though this is not always possible)
a way for the producer to embed versioning information to the serialized data
a backward compatible strategy to handle unknown fields
In a web API, generally the API version that the consumer accepts is embedded in the Accept header (e.g. Accept: application/vnd.myapp-v1+json application/vnd.myapp-v2+json means the consumer can handle either version 1 and version 2 of your API) or less commonly in the URL (e.g. https://api.twitter.com/1/statuses/user_timeline.json). This is generally used for major versions (i.e. backward incompatible changes). If the server and the client does not have a matching Accept header, then the communication fails (or proceeds in best-effort basis or fallback to a default baseline protocol, depending on the nature of the application).
The producer then generates a serialized data in one of the requested version, then embed this version info into the serialized data (e.g. as a field named version). The consumer should use the version information embedded in the data to determine how to parse the serialized data. The version information in the data should also contain minor version (i.e. for backward compatible changes), generally consumers should be able to ignore the minor version information and still process the data correctly although understanding the minor version may allow the client to make additional assumptions about how the data should be processed.
A common strategy to handle unknown fields is like how HTML and CSS are parsed. When the consumer sees an unknown fields they should ignore it, and when the data is missing a field that the client is expecting, it should use a default value; depending on the nature of the communication, you may also want to specify some fields that are mandatory (i.e. missing fields is considered fatal error). Fields added within minor versions should always be optional field; minor version can add optional fields or change fields semantic as long as it's backward compatible, while major version can delete fields or add mandatory fields or change fields semantic in a backward incompatible manner.
In an extensible serialization format (like JSON or XML), the data should be self-descriptive, in other words, the field names should always be stored together with the data; you should not rely on the specific data being available on specific positions.
Don't use DataContractJsonSerializer, as the name says, the objects that are processed through this class will have to:
a) Be marked with [DataContract] and [DataMember] attributes.
b) Be strictly compliant with the defined "Contract" that is, nothing less and nothing more that it is defined. Any extra or missing [DataMember] will make the deserialization to throw an exception.
If you want to be flexible enough, then use the JavaScriptSerializer if you want to go for the cheap option... or use this library:
http://json.codeplex.com/
This will give you enough control over your JSON serialization.
Imagine you have an object in its early days.
public class Customer
{
public string Name;
public string LastName;
}
Once serialized it will look like this:
{ Name: "John", LastName: "Doe" }
If you change your object definition to add / remove fields. The deserialization will occur smoothly if you use, for example, JavaScriptSerializer.
public class Customer
{
public string Name;
public string LastName;
public int Age;
}
If yo try to de-serialize the last json to this new class, no error will be thrown. The thing is that your new fields will be set to their defaults. In this example: "Age" will be set to zero.
You can include, in your own conventions, a field present in all your objects, that contains the version number. In this case you can tell the difference between an empty field or a version inconsistence.
So lets say: You have your class Customer v1 serialized:
{ Version: 1, LastName: "Doe", Name: "John" }
You want to deserialize into a Customer v2 instance, you will have:
{ Version: 1, LastName: "Doe", Name: "John", Age: 0}
You can somehow, detect what fields in your object are somehow reliable and what's not. In this case you know that your v2 object instance is coming from a v1 object instance, so the field Age should not be trusted.
I have in mind that you should use also a custom attribute, e.g. "MinVersion", and mark each field with the minimum supported version number, so you get something like this:
public class Customer
{
[MinVersion(1)]
public int Version;
[MinVersion(1)]
public string Name;
[MinVersion(1)]
public string LastName;
[MinVersion(2)]
public int Age;
}
Then later you can access this meta-data and do whatever you might need with that.

What are better ways to create a method that takes many arguments? (10+?)

I was looking at some code of a fellow developer, and almost cried. In the method definition there are 12 arguments. From my experience..this isn't good. If it were me, I would have sent in an object of some sort.
Is there another / more preferred way to do this (in other words, what's the best way to fix this and explain why)?
public long Save (
String today,
String name,
String desc,
int ID,
String otherNm,
DateTime dt,
int status,
String periodID,
String otherDt,
String submittedDt
)
ignore my poor variable names - they are examples
It highly depends on the language.
In a language without compile-time typechecking (e.g. python, javascript, etc.) you should use keyword arguments (common in python: you can access them like a dictionary passed in as an argument) or objects/dictionaries you manually pass in as arguments (common in javascript).
However the "argument hell" you described is sometimes "the right way to do things" for certain languages with compile-time typechecking, because using objects will obfuscate the semantics from the typechecker. The solution then would be to use a better language with compile-time typechecking which allows pattern-matching of objects as arguments.
Yes, use objects. Also, the function is probably doing too much if it needs all of this information, so use smaller functions.
Use objects.
class User { ... }
User user = ...
Save(user);
It decision provides easy way for adding new parameters.
It depends on how complex the function is. If it does something non-trivial with each of those arguments, it should probably be split. If it just passes them through, they should probably be collected in an object. But if it just creates a row in a table, it's not really big deal. It's less of a deal if your language supports keyword arguments.
I imagine the issue you're experiencing is being able to look at the method call and know what argument is receiving what value. This is a pernicious problem in a language like Java, which lacks something like keyword arguments or JSON hashes to pass named arguments.
In this situation, the Builder pattern is a useful solution. It's more objects, three total, but leads to more comprehensible code for the problem you're describing. So the three objects in this case would be as such:
Thing: stateful entity, typically immutable (i.e. getters only)
ThingBuilder: factory class, creates a Thing entity and sets its values.
ThingDAO: not necessary for using the Builder pattern, but addresses your question.
Interaction
/*
ThingBuilder is a static inner class of Thing, where each of its
"set" method calls returns the ThingBuilder instance being worked with
while the final "build()" call returns the instantiated Thing instance.
*/
Thing thing = Thing.createBuilder().
.setToday("2012/04/01")
.setName("Example")
// ...etc...
.build();
// the Thing instance as get methods for each property
thing.getName();
// get your reference to thingDAO however it's done
thingDAO.save(thing);
The result is you get named arguments and an immutable instance.

Are hard-coded STRINGS ever acceptable?

Similar to Is hard-coding literals ever acceptable?, but I'm specifically thinking of "magic strings" here.
On a large project, we have a table of configuration options like these:
Name Value
---- -----
FOO_ENABLED Y
BAR_ENABLED N
...
(Hundreds of them).
The common practice is to call a generic function to test an option like this:
if (config_options.value('FOO_ENABLED') == 'Y') ...
(Of course, this same option may need to be checked in many places in the system code.)
When adding a new option, I was considering adding a function to hide the "magic string" like this:
if (config_options.foo_enabled()) ...
However, colleagues thought I'd gone overboard and objected to doing this, preferring the hard-coding because:
That's what we normally do
It makes it easier to see what's going on when debugging the code
The trouble is, I can see their point! Realistically, we are never going to rename the options for any reason, so about the only advantage I can think of for my function is that the compiler would catch any typo like fo_enabled(), but not 'FO_ENABLED'.
What do you think? Have I missed any other advantages/disadvantages?
If I use a string once in the code, I don't generally worry about making it a constant somewhere.
If I use a string twice in the code, I'll consider making it a constant.
If I use a string three times in the code, I'll almost certainly make it a constant.
if (config_options.isTrue('FOO_ENABLED')) {...
}
Restrict your hard coded Y check to one place, even if it means writing a wrapper class for your Map.
if (config_options.isFooEnabled()) {...
}
Might seem okay until you have 100 configuration options and 100 methods (so here you can make a judgement about future application growth and needs before deciding on your implementation). Otherwise it is better to have a class of static strings for parameter names.
if (config_options.isTrue(ConfigKeys.FOO_ENABLED)) {...
}
I realise the question is old, but it came up on my margin.
AFAIC, the issue here has not been identified accurately, either in the question, or the answers. Forget about 'harcoding strings" or not, for a moment.
The database has a Reference table, containing config_options. The PK is a string.
There are two types of PKs:
Meaningful Identifiers, that the users (and developers) see and use. These PKs are supposed to be stable, they can be relied upon.
Meaningless Id columns which the users should never see, that the developers have to be aware of, and code around. These cannot be relied upon.
It is ordinary, normal, to write code using the absolute value of a meaningful PK IF CustomerCode = "IBM" ... or IF CountryCode = "AUS" etc.
referencing the absolute value of a meaningless PK is not acceptable (due to auto-increment; gaps being changed; values being replaced wholesale).
.
Your reference table uses meaningful PKs. Referencing those literal strings in code is unavoidable. Hiding the value will make maintenance more difficult; the code is no longer literal; your colleagues are right. Plus there is the additional redundant function that chews cycles. If there is a typo in the literal, you will soon find that out during Dev testing, long before UAT.
hundreds of functions for hundreds of literals is absurd. If you do implement a function, then Normalise your code, and provide a single function that can be used for any of the hundreds of literals. In which case, we are back to a naked literal, and the function can be dispensed with.
the point is, the attempt to hide the literal has no value.
.
It cannot be construed as "hardcoding", that is something quite different. I think that is where your issue is, identifying these constructs as "hardcoded". It is just referencing a Meaningfull PK literally.
Now from the perspective of any code segment only, if you use the same value a few times, you can improve the code by capturing the literal string in a variable, and then using the variable in the rest of the code block. Certainly not a function. But that is an efficiency and good practice issue. Even that does not change the effect IF CountryCode = #cc_aus
I really should use constants and no hard coded literals.
You can say they won't be changed, but you may never know. And it is best to make it a habit. To use symbolic constants.
In my experience, this kind of issue is masking a deeper problem: failure to do actual OOP and to follow the DRY principle.
In a nutshell, capture the decision at startup time by an appropriate definition for each action inside the if statements, and then throw away both the config_options and the run-time tests.
Details below.
The sample usage was:
if (config_options.value('FOO_ENABLED') == 'Y') ...
which raises the obvious question, "What's going on in the ellipsis?", especially given the following statement:
(Of course, this same option may need to be checked in many places in the system code.)
Let's assume that each of these config_option values really does correspond to a single problem domain (or implementation strategy) concept.
Instead of doing this (repeatedly, in various places throughout the code):
Take a string (tag),
Find its corresponding other string (value),
Test that value as a boolean-equivalent,
Based on that test, decide whether to perform some action.
I suggest encapsulating the concept of a "configurable action".
Let's take as an example (obviously just as hypthetical as FOO_ENABLED ... ;-) that your code has to work in either English units or metric units. If METRIC_ENABLED is "true", convert user-entered data from metric to English for internal computation, and convert back prior to displaying results.
Define an interface:
public interface MetricConverter {
double toInches(double length);
double toCentimeters(double length);
double toPounds(double weight);
double toKilograms(double weight);
}
which identifies in one place all the behavior associated with the concept of METRIC_ENABLED.
Then write concrete implementations of all the ways those behaviors are to be carried out:
public class NullConv implements MetricConverter {
double toInches(double length) {return length;}
double toCentimeters(double length) {return length;}
double toPounds(double weight) {return weight;}
double toKilograms(double weight) {return weight;}
}
and
// lame implementation, just for illustration!!!!
public class MetricConv implements MetricConverter {
public static final double LBS_PER_KG = 2.2D;
public static final double CM_PER_IN = 2.54D
double toInches(double length) {return length * CM_PER_IN;}
double toCentimeters(double length) {return length / CM_PER_IN;}
double toPounds(double weight) {return weight * LBS_PER_KG;}
double toKilograms(double weight) {return weight / LBS_PER_KG;}
}
At startup time, instead of loading a bunch of config_options values, initialize a set of configurable actions, as in:
MetricConverter converter = (metricOption()) ? new MetricConv() : new NullConv();
(where the expression metricOption() above is a stand-in for whatever one-time-only check you need to make, including looking at the value of METRIC_ENABLED ;-)
Then, wherever the code would have said:
double length = getLengthFromGui();
if (config_options.value('METRIC_ENABLED') == 'Y') {
length = length / 2.54D;
}
// do some computation to produce result
// ...
if (config_options.value('METRIC_ENABLED') == 'Y') {
result = result * 2.54D;
}
displayResultingLengthOnGui(result);
rewrite it as:
double length = converter.toInches(getLengthFromGui());
// do some computation to produce result
// ...
displayResultingLengthOnGui(converter.toCentimeters(result));
Because all of the implementation details related to that one concept are now packaged cleanly, all future maintenance related to METRIC_ENABLED can be done in one place. In addition, the run-time trade-off is a win; the "overhead" of invoking a method is trivial compared with the overhead of fetching a String value from a Map and performing String#equals.
I believe that the two reasons you have mentioned, Possible misspelling in string, that cannot be detected until run time and the possibility (although slim) of a name change would justify your idea.
On top of that you can get typed functions, now it seems you only store booleans, what if you need to store an int, a string etc. I would rather use get_foo() with a type, than get_string("FOO") or get_int("FOO").
I think there are two different issues here:
In the current project, the convention of using hard-coded strings is already well established, so all the developers working on the project are familiar with it. It might be a sub-optimal convention for all the reasons that have been listed, but everybody familiar with the code can look at it and instinctively knows what the code is supposed to do. Changing the code so that in certain parts, it uses the "new" functionality will make the code slightly harder to read (because people will have to think and remember what the new convention does) and thus a little harder to maintain. But I would guess that changing over the whole project to the new convention would potentially be prohibitively expensive unless you can quickly script the conversion.
On a new project, symbolic constants are the way IMO, for all the reasons listed. Especially because anything that makes the compiler catch errors at compile time that would otherwise be caught by a human at run time is a very useful convention to establish.
Another thing to consider is intent. If you are on a project that requires localization hard coded strings can be ambiguous. Consider the following:
const string HELLO_WORLD = "Hello world!";
print(HELLO_WORLD);
The programmer's intent is clear. Using a constant implies that this string does not need to be localized. Now look at this example:
print("Hello world!");
Here we aren't so sure. Did the programmer really not want this string to be localized or did the programmer forget about localization while he was writing this code?
I too prefer a strongly-typed configuration class if it is used through-out the code. With properly named methods you don't lose any readability. If you need to do conversions from strings to another data type (decimal/float/int), you don't need to repeat the code that does the conversion in multiple places and can cache the result so the conversion only takes place once. You've already got the basis of this in place already so I don't think it would take much to get used to the new way of doing things.

What is Type-safe?

What does "type-safe" mean?
Type safety means that the compiler will validate types while compiling, and throw an error if you try to assign the wrong type to a variable.
Some simple examples:
// Fails, Trying to put an integer in a string
String one = 1;
// Also fails.
int foo = "bar";
This also applies to method arguments, since you are passing explicit types to them:
int AddTwoNumbers(int a, int b)
{
return a + b;
}
If I tried to call that using:
int Sum = AddTwoNumbers(5, "5");
The compiler would throw an error, because I am passing a string ("5"), and it is expecting an integer.
In a loosely typed language, such as javascript, I can do the following:
function AddTwoNumbers(a, b)
{
return a + b;
}
if I call it like this:
Sum = AddTwoNumbers(5, "5");
Javascript automaticly converts the 5 to a string, and returns "55". This is due to javascript using the + sign for string concatenation. To make it type-aware, you would need to do something like:
function AddTwoNumbers(a, b)
{
return Number(a) + Number(b);
}
Or, possibly:
function AddOnlyTwoNumbers(a, b)
{
if (isNaN(a) || isNaN(b))
return false;
return Number(a) + Number(b);
}
if I call it like this:
Sum = AddTwoNumbers(5, " dogs");
Javascript automatically converts the 5 to a string, and appends them, to return "5 dogs".
Not all dynamic languages are as forgiving as javascript (In fact a dynamic language does not implicity imply a loose typed language (see Python)), some of them will actually give you a runtime error on invalid type casting.
While its convenient, it opens you up to a lot of errors that can be easily missed, and only identified by testing the running program. Personally, I prefer to have my compiler tell me if I made that mistake.
Now, back to C#...
C# supports a language feature called covariance, this basically means that you can substitute a base type for a child type and not cause an error, for example:
public class Foo : Bar
{
}
Here, I created a new class (Foo) that subclasses Bar. I can now create a method:
void DoSomething(Bar myBar)
And call it using either a Foo, or a Bar as an argument, both will work without causing an error. This works because C# knows that any child class of Bar will implement the interface of Bar.
However, you cannot do the inverse:
void DoSomething(Foo myFoo)
In this situation, I cannot pass Bar to this method, because the compiler does not know that Bar implements Foo's interface. This is because a child class can (and usually will) be much different than the parent class.
Of course, now I've gone way off the deep end and beyond the scope of the original question, but its all good stuff to know :)
Type-safety should not be confused with static / dynamic typing or strong / weak typing.
A type-safe language is one where the only operations that one can execute on data are the ones that are condoned by the data's type. That is, if your data is of type X and X doesn't support operation y, then the language will not allow you to to execute y(X).
This definition doesn't set rules on when this is checked. It can be at compile time (static typing) or at runtime (dynamic typing), typically through exceptions. It can be a bit of both: some statically typed languages allow you to cast data from one type to another, and the validity of casts must be checked at runtime (imagine that you're trying to cast an Object to a Consumer - the compiler has no way of knowing whether it's acceptable or not).
Type-safety does not necessarily mean strongly typed, either - some languages are notoriously weakly typed, but still arguably type safe. Take Javascript, for example: its type system is as weak as they come, but still strictly defined. It allows automatic casting of data (say, strings to ints), but within well defined rules. There is to my knowledge no case where a Javascript program will behave in an undefined fashion, and if you're clever enough (I'm not), you should be able to predict what will happen when reading Javascript code.
An example of a type-unsafe programming language is C: reading / writing an array value outside of the array's bounds has an undefined behaviour by specification. It's impossible to predict what will happen. C is a language that has a type system, but is not type safe.
Type safety is not just a compile time constraint, but a run time constraint. I feel even after all this time, we can add further clarity to this.
There are 2 main issues related to type safety. Memory** and data type (with its corresponding operations).
Memory**
A char typically requires 1 byte per character, or 8 bits (depends on language, Java and C# store unicode chars which require 16 bits).
An int requires 4 bytes, or 32 bits (usually).
Visually:
char: |-|-|-|-|-|-|-|-|
int : |-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-|
A type safe language does not allow an int to be inserted into a char at run-time (this should throw some kind of class cast or out of memory exception). However, in a type unsafe language, you would overwrite existing data in 3 more adjacent bytes of memory.
int >> char:
|-|-|-|-|-|-|-|-| |?|?|?|?|?|?|?|?| |?|?|?|?|?|?|?|?| |?|?|?|?|?|?|?|?|
In the above case, the 3 bytes to the right are overwritten, so any pointers to that memory (say 3 consecutive chars) which expect to get a predictable char value will now have garbage. This causes undefined behavior in your program (or worse, possibly in other programs depending on how the OS allocates memory - very unlikely these days).
** While this first issue is not technically about data type, type safe languages address it inherently and it visually describes the issue to those unaware of how memory allocation "looks".
Data Type
The more subtle and direct type issue is where two data types use the same memory allocation. Take a int vs an unsigned int. Both are 32 bits. (Just as easily could be a char[4] and an int, but the more common issue is uint vs. int).
|-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-|
|-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-| |-|-|-|-|-|-|-|-|
A type unsafe language allows the programmer to reference a properly allocated span of 32 bits, but when the value of a unsigned int is read into the space of an int (or vice versa), we again have undefined behavior. Imagine the problems this could cause in a banking program:
"Dude! I overdrafted $30 and now I have $65,506 left!!"
...'course, banking programs use much larger data types. ;) LOL!
As others have already pointed out, the next issue is computational operations on types. That has already been sufficiently covered.
Speed vs Safety
Most programmers today never need to worry about such things unless they are using something like C or C++. Both of these languages allow programmers to easily violate type safety at run time (direct memory referencing) despite the compilers' best efforts to minimize the risk. HOWEVER, this is not all bad.
One reason these languages are so computationally fast is they are not burdened by verifying type compatibility during run time operations like, for example, Java. They assume the developer is a good rational being who won't add a string and an int together and for that, the developer is rewarded with speed/efficiency.
Many answers here conflate type-safety with static-typing and dynamic-typing. A dynamically typed language (like smalltalk) can be type-safe as well.
A short answer: a language is considered type-safe if no operation leads to undefined behavior. Many consider the requirement of explicit type conversions necessary for a language to be strictly typed, as automatic conversions can sometimes leads to well defined but unexpected/unintuitive behaviors.
A programming language that is 'type-safe' means following things:
You can't read from uninitialized variables
You can't index arrays beyond their bounds
You can't perform unchecked type casts
An explanation from a liberal arts major, not a comp sci major:
When people say that a language or language feature is type safe, they mean that the language will help prevent you from, for example, passing something that isn't an integer to some logic that expects an integer.
For example, in C#, I define a function as:
void foo(int arg)
The compiler will then stop me from doing this:
// call foo
foo("hello world")
In other languages, the compiler would not stop me (or there is no compiler...), so the string would be passed to the logic and then probably something bad will happen.
Type safe languages try to catch more at "compile time".
On the down side, with type safe languages, when you have a string like "123" and you want to operate on it like an int, you have to write more code to convert the string to an int, or when you have an int like 123 and want to use it in a message like, "The answer is 123", you have to write more code to convert/cast it to a string.
To get a better understanding do watch the below video which demonstrates code in type safe language (C#) and NOT type safe language ( javascript).
http://www.youtube.com/watch?v=Rlw_njQhkxw
Now for the long text.
Type safety means preventing type errors. Type error occurs when data type of one type is assigned to other type UNKNOWINGLY and we get undesirable results.
For instance JavaScript is a NOT a type safe language. In the below code “num” is a numeric variable and “str” is string. Javascript allows me to do “num + str” , now GUESS will it do arithmetic or concatenation .
Now for the below code the results are “55” but the important point is the confusion created what kind of operation it will do.
This is happening because javascript is not a type safe language. Its allowing to set one type of data to the other type without restrictions.
<script>
var num = 5; // numeric
var str = "5"; // string
var z = num + str; // arthimetic or concat ????
alert(z); // displays “55”
</script>
C# is a type safe language. It does not allow one data type to be assigned to other data type. The below code does not allow “+” operator on different data types.
Concept:
To be very simple Type Safe like the meanings, it makes sure that type of the variable should be safe like
no wrong data type e.g. can't save or initialized a variable of string type with integer
Out of bound indexes are not accessible
Allow only the specific memory location
so it is all about the safety of the types of your storage in terms of variables.
Type-safe means that programmatically, the type of data for a variable, return value, or argument must fit within a certain criteria.
In practice, this means that 7 (an integer type) is different from "7" (a quoted character of string type).
PHP, Javascript and other dynamic scripting languages are usually weakly-typed, in that they will convert a (string) "7" to an (integer) 7 if you try to add "7" + 3, although sometimes you have to do this explicitly (and Javascript uses the "+" character for concatenation).
C/C++/Java will not understand that, or will concatenate the result into "73" instead. Type-safety prevents these types of bugs in code by making the type requirement explicit.
Type-safety is very useful. The solution to the above "7" + 3 would be to type cast (int) "7" + 3 (equals 10).
Try this explanation on...
TypeSafe means that variables are statically checked for appropriate assignment at compile time. For example, consder a string or an integer. These two different data types cannot be cross-assigned (ie, you can't assign an integer to a string nor can you assign a string to an integer).
For non-typesafe behavior, consider this:
object x = 89;
int y;
if you attempt to do this:
y = x;
the compiler throws an error that says it can't convert a System.Object to an Integer. You need to do that explicitly. One way would be:
y = Convert.ToInt32( x );
The assignment above is not typesafe. A typesafe assignement is where the types can directly be assigned to each other.
Non typesafe collections abound in ASP.NET (eg, the application, session, and viewstate collections). The good news about these collections is that (minimizing multiple server state management considerations) you can put pretty much any data type in any of the three collections. The bad news: because these collections aren't typesafe, you'll need to cast the values appropriately when you fetch them back out.
For example:
Session[ "x" ] = 34;
works fine. But to assign the integer value back, you'll need to:
int i = Convert.ToInt32( Session[ "x" ] );
Read about generics for ways that facility helps you easily implement typesafe collections.
C# is a typesafe language but watch for articles about C# 4.0; interesting dynamic possibilities loom (is it a good thing that C# is essentially getting Option Strict: Off... we'll see).
Type-Safe is code that accesses only the memory locations it is authorized to access, and only in well-defined, allowable ways.
Type-safe code cannot perform an operation on an object that is invalid for that object. The C# and VB.NET language compilers always produce type-safe code, which is verified to be type-safe during JIT compilation.
Type-safe means that the set of values that may be assigned to a program variable must fit well-defined and testable criteria. Type-safe variables lead to more robust programs because the algorithms that manipulate the variables can trust that the variable will only take one of a well-defined set of values. Keeping this trust ensures the integrity and quality of the data and the program.
For many variables, the set of values that may be assigned to a variable is defined at the time the program is written. For example, a variable called "colour" may be allowed to take on the values "red", "green", or "blue" and never any other values. For other variables those criteria may change at run-time. For example, a variable called "colour" may only be allowed to take on values in the "name" column of a "Colours" table in a relational database, where "red, "green", and "blue", are three values for "name" in the "Colours" table, but some other part of the computer program may be able to add to that list while the program is running, and the variable can take on the new values after they are added to the Colours table.
Many type-safe languages give the illusion of "type-safety" by insisting on strictly defining types for variables and only allowing a variable to be assigned values of the same "type". There are a couple of problems with this approach. For example, a program may have a variable "yearOfBirth" which is the year a person was born, and it is tempting to type-cast it as a short integer. However, it is not a short integer. This year, it is a number that is less than 2009 and greater than -10000. However, this set grows by 1 every year as the program runs. Making this a "short int" is not adequate. What is needed to make this variable type-safe is a run-time validation function that ensures that the number is always greater than -10000 and less than the next calendar year. There is no compiler that can enforce such criteria because these criteria are always unique characteristics of the problem domain.
Languages that use dynamic typing (or duck-typing, or manifest typing) such as Perl, Python, Ruby, SQLite, and Lua don't have the notion of typed variables. This forces the programmer to write a run-time validation routine for every variable to ensure that it is correct, or endure the consequences of unexplained run-time exceptions. In my experience, programmers in statically typed languages such as C, C++, Java, and C# are often lulled into thinking that statically defined types is all they need to do to get the benefits of type-safety. This is simply not true for many useful computer programs, and it is hard to predict if it is true for any particular computer program.
The long & the short.... Do you want type-safety? If so, then write run-time functions to ensure that when a variable is assigned a value, it conforms to well-defined criteria. The down-side is that it makes domain analysis really difficult for most computer programs because you have to explicitly define the criteria for each program variable.
Type Safety
In modern C++, type safety is very important. Type safety means that you use the types correctly and, therefore, avoid unsafe casts and unions. Every object in C++ is used according to its type and an object needs to be initialized before its use.
Safe Initialization: {}
The compiler protects from information loss during type conversion. For example,
int a{7}; The initialization is OK
int b{7.5} Compiler shows ERROR because of information loss.\
Unsafe Initialization: = or ()
The compiler doesn't protect from information loss during type conversion.
int a = 7 The initialization is OK
int a = 7.5 The initialization is OK, but information loss occurs. The actual value of a will become 7.0
int c(7) The initialization is OK
int c(7.5) The initialization is OK, but information loss occurs. The actual value of a will become 7.0

api documentation and "value limits": do they match?

Do you often see in API documentation (as in 'javadoc of public functions' for example) the description of "value limits" as well as the classic documentation ?
Note: I am not talking about comments within the code
By "value limits", I mean:
does a parameter can support a null value (or an empty String, or...) ?
does a 'return value' can be null or is guaranteed to never be null (or can be "empty", or...) ?
Sample:
What I often see (without having access to source code) is:
/**
* Get all readers name for this current Report. <br />
* <b>Warning</b>The Report must have been published first.
* #param aReaderNameRegexp filter in order to return only reader matching the regexp
* #return array of reader names
*/
String[] getReaderNames(final String aReaderNameRegexp);
What I like to see would be:
/**
* Get all readers name for this current Report. <br />
* <b>Warning</b>The Report must have been published first.
* #param aReaderNameRegexp filter in order to return only reader matching the regexp
* (can be null or empty)
* #return array of reader names
* (null if Report has not yet been published,
* empty array if no reader match criteria,
* reader names array matching regexp, or all readers if regexp is null or empty)
*/
String[] getReaderNames(final String aReaderNameRegexp);
My point is:
When I use a library with a getReaderNames() function in it, I often do not even need to read the API documentation to guess what it does. But I need to be sure how to use it.
My only concern when I want to use this function is: what should I expect in term of parameters and return values ? That is all I need to know to safely setup my parameters and safely test the return value, yet I almost never see that kind of information in API documentation...
Edit:
This can influence the usage or not for checked or unchecked exceptions.
What do you think ? value limits and API, do they belong together or not ?
I think they can belong together but don't necessarily have to belong together. In your scenario, it seems like it makes sense that the limits are documented in such a way that they appear in the generated API documentation and intellisense (if the language/IDE support it).
I think it does depend on the language as well. For example, Ada has a native data type that is a "restricted integer", where you define an integer variable and explicitly indicate that it will only (and always) be within a certain numeric range. In that case, the datatype itself indicates the restriction. It should still be visible and discoverable through the API documentation and intellisense, but wouldn't be something that a developer has to specify in the comments.
However, languages like Java and C# don't have this type of restricted integer, so the developer would have to specify it in the comments if it were information that should become part of the public documentation.
I think those kinds of boundary conditions most definitely belong in the API. However, I would (and often do) go a step further and indicate WHAT those null values mean. Either I indicate it will throw an exception, or I explain what the expected results are when the boundary value is passed in.
It's hard to remember to always do this, but it's a good thing for users of your class. It's also difficult to maintain it if the contract the method presents changes (like null values are changed to no be allowed)... you have to be diligent also to update the docs when you change the semantics of the method.
Question 1
Do you often see in API documentation (as in 'javadoc of public functions' for example) the description of "value limits" as well as the classic documentation?
Almost never.
Question 2
My only concern when I want to use this function is: what should I expect in term of parameters and return values ? That is all I need to know to safely setup my parameters and safely test the return value, yet I almost never see that kind of information in API documentation...
If I used a function not properly I would expect a RuntimeException thrown by the method or a RuntimeException in another (sometimes very far) part of the program.
Comments like #param aReaderNameRegexp filter in order to ... (can be null or empty) seems to me a way to implement Design by Contract in a human-being language inside Javadoc.
Using Javadoc to enforce Design by Contract was used by iContract, now resurrected into JcontractS, that let you specify invariants, preconditions, postconditions, in more formalized way compared to the human-being language.
Question 3
This can influence the usage or not for checked or unchecked exceptions.
What do you think ? value limits and API, do they belong together or not ?
Java language doesn't have a Design by Contract feature, so you might be tempted to use Execption but I agree with you about the fact that you have to be aware about When to choose checked and unchecked exceptions. Probably you might use unchecked IllegalArgumentException, IllegalStateException, or you might use unit testing, but the major problem is how to communicate to other programmers that such code is about Design By Contract and should be considered as a contract before changing it too lightly.
I think they do, and have always placed comments in the header files (c++) arcordingly.
In addition to valid input/output/return comments, I also note which exceptions are likly to be thrown by the function (since I often want to use the return value for...well returning a value, I prefer exceptions over error codes)
//File:
// Should be a path to the teexture file to load, if it is not a full path (eg "c:\example.png") it will attempt to find the file usign the paths provided by the DataSearchPath list
//Return: The pointer to a Texture instance is returned, in the event of an error, an exception is thrown. When you are finished with the texture you chould call the Free() method.
//Exceptions:
//except::FileNotFound
//except::InvalidFile
//except::InvalidParams
//except::CreationFailed
Texture *GetTexture(const std::string &File);
#Fire Lancer: Right! I forgot about exception, but I would like to see them mentioned, especially the unchecked 'runtime' exception that this public method could throw
#Mike Stone:
you have to be diligent also to update the docs when you change the semantics of the method.
Mmmm I sure hope that the public API documentation is at the very least updated whenever a change -- that affects the contract of the function -- takes place. If not, those API documentations could be drop altogether.
To add food to yours thoughts (and go with #Scott Dorman), I just stumble upon the future of java7 annotations
What does that means ? That certain 'boundary conditions', rather than being in the documentation, should be better off in the API itself, and automatically used, at compilation time, with appropriate 'assert' generated code.
That way, if a '#CheckForNull' is in the API, the writer of the function might get away with not even documenting it! And if the semantic change, its API will reflect that change (like 'no more #CheckForNull' for instance)
That kind of approach suggests that documentation, for 'boundary conditions', is an extra bonus rather than a mandatory practice.
However, that does not cover the special values of the return object of a function. For that, a complete documentation is still needed.