I have encountered this topic lately and couldn't understand why they are needed.
Can you explain why I should use them in my projects and how they can ease my life.
Examples will be great, and where from I can learn this topic little more.
At least you have framed the question from the correct perspective =)
The usual reasons for using a code generator are given as productivity and consistency because they assume that the solution to a consistent and repetitive problem is to throw more code at it. I would argue that any time you are considering code generation, look at why you are generating code and see if you can solve the problem through other means.
A classic example of this is data access; you could generate 250 classes ( 1 for each table in the schema ) effectively creating a table gateway solution, or you could build something more like a domain model and use NHibernate / ActiveRecord / LightSpeed / [pick your orm] to map a rich domain model onto the database.
While both the hand rolled solution and ORM are effectively code generators, the primary difference is when the code is generated. With the ORM it is an implicit step that happens at run-time and therefore is one-way by it's nature. The hand rolled solution requires and explicit step to generate the code during development and the likelihood that the generated classes will need customising at some point therefore creating problems when you re-generate the code. The explicit step that must happen during development introduces friction into the development process and often leads to code that violates DRY ( although some argue that generated code can never violate DRY ).
Another reason for touting code generation comes from the MDA / MDE world ( Model Driven Architecture / Engineering ). I don't put much stock in this but rather than providing a number of poorly expressed arguments, I'm simply going to co-opt someone elses - http://www.infoq.com/articles/8-reasons-why-MDE-fails.
IMHO code generation is the only solution in an exceedingly narrow set of problems and whenever you are considering it, you should probably take a second look at the real problem you are trying to solve and see if there is a better solution.
One type of code generation that really does enhance productivity is "micro code-generation" where the use of macros and templates allow a developer to generate new code directly in the IDE and tab / type their way through placeholders (eg namespace / classname etc). This sort of code generation is a feature of resharper and I use it heavily every day. The reason that micro-generation benefits where most large scale code generation fails is that the generated code is not tied back to any other resource that must be kept in sync and therefore once the code is generated, it is just like all the other code in the solution.
#John
Moving the creation of "basic classes" from the IDE into xml / dsl is often seen when doing big bang development - a classic example would be developers try to reverse engineer the database into a domain model. Unless the code generator is very well written it simply introduces an additional burden on the developer in that every time they need to update the domain model, they either have to context-switch and update the xml / dsl or they have to extend the domain model and then port those changes back to the xml / dsl ( effectively doing the work twice).
There are some code generators that work very well in this space ( the LightSpeed designer is the only one I can think of atm ) by acting as the engine for a design surface but often
these code generators generate terrible code that cannot be maintained (eg winforms / webforms design surfaces, EF1 design surface) and therefore rapidly undo any productivity benefits gained from using the code generator in the first place.
Well, it's either:
you write 250 classes, all pretty much the same, but slightly different, e.g. to do data access; takes you a week, and it's boring and error-prone and annoying
OR:
you invest 30 minutes into generating a code template, and let a generation engine handle the grunt work in another 30 minutes
So a code generator gives you:
speed
reproducability
a lot less errors
a lot more free time! :-)
Excellent examples:
Linq-to-SQL T4 templates by Damien Guard to generate one separate file per class in your database model, using the best kept Visual Studio 2008 secret - T4 templates
PLINQO - same thing, but for Codesmith's generator
and countless more.....
Anytime you need to produce large amounts of repetetive boilerplate code, the code generator is the guy for the job. Last time I used a code generator was when creating a custom Data Access Layer for a project, where the skeleton for various CRUD actions was created based on an object model. Instead of hand-coding all those classes, I put together a template-driven code generator (using StringTemplate) to make it for me. The advandages of this procedure was:
It was faster (there was a large amount of code to generate)
I could regenerate the code in a whim in case I detected a bug (code can sometimes have bugs in the early versions)
Less error prone; when we had an error in the generated code it was everywhere which means that it was more likely to be found (and, as noted in the previous point, it was easy to fix it and regenerate the code).
Using GUI builders, that will generate code for you is a common practice. Thanks to this you don't need to manually create all widgets. You just drag&drop them and the use generated code. For simple widgets this really saves time (I have used this a lot for wxWidgets).
Really, when you are using almost any programming language, you are using a "code generator" (except for assembly or machine code.) I often write little 200-line scripts that crank out a few thousand lines of C. There is also software you can get which helps generate certain types of code (yacc and lex, for example, are used to generate parsers to create programming languages.)
The key here is to think of your code generator's input as the actual source code, and think of the stuff it spits out as just part of the build process. In which case, you are writing in a higher-level language with fewer actual lines of code to deal with.
For example, here is a very long and tedious file I (didn't) write as part of my work modifying the Quake2-based game engine CRX. It takes the integer values of all #defined constants from two of the headers, and makes them into "cvars" (variables for the in-game console.)
http://meliaserlow.dyndns.tv:8000/alienarena/lua_source/game/cvar_constants.c
Here is the short Bash script which generated that code at compile-time:
http://meliaserlow.dyndns.tv:8000/alienarena/lua_source/autogen/constant_cvars.sh
Now, which would you rather maintain? They are both equivalent in terms of what they describe, but one is vastly longer and more annoying to deal with.
The canonical example of this is data access, but I have another example. I've worked on a messaging system that communicates over serial port, sockets, etc., and I found I kept having to write classes like this over and over again:
public class FooMessage
{
public FooMessage()
{
}
public FooMessage(int bar, string baz, DateTime blah)
{
this.Bar = bar;
this.Baz = baz;
this.Blah = blah;
}
public void Read(BinaryReader reader)
{
this.Bar = reader.ReadInt32();
this.Baz = Encoding.ASCII.GetString(reader.ReadBytes(30));
this.Blah = new DateTime(reader.ReadInt16(), reader.ReadByte(),
reader.ReadByte());
}
public void Write(BinaryWriter writer)
{
writer.Write(this.Bar);
writer.Write(Encoding.ASCII.GetBytes(
this.Baz.PadRight(30).Substring(0, 30)));
writer.Write((Int16)this.Blah.Year);
writer.Write((byte)this.Blah.Month);
writer.Write((byte)this.Blah.Day);
}
public int Bar { get; set; }
public string Baz { get; set; }
public DateTime Blah { get; set; }
}
Try to imagine, if you will, writing this code for no fewer than 300 different types of messages. The same boring, tedious, error-prone code being written, over and over again. I managed to write about 3 of these before I decided it would be easier for me to just write a code generator, so I did.
I won't post the code-gen code, it's a lot of arcane CodeDom stuff, but the bottom line is that I was able to compact the entire system down to a single XML file:
<Messages>
<Message ID="12345" Name="Foo">
<ByteField Name="Bar"/>
<TextField Name="Baz" Length="30"/>
<DateTimeField Name="Blah" Precision="Day"/>
</Message>
(More messages)
</Messages>
How much easier is this? (Rhetorical question.) I could finally breathe. I even added some bells and whistles so it was able to generate a "proxy", and I could write code like this:
var p = new MyMessagingProtocol(...);
SetFooResult result = p.SetFoo(3, "Hello", DateTime.Today);
In the end I'd say this saved me writing a good 7500 lines of code and turned a 3-week task into a 3-day task (well, plus the couple of days required to write the code-gen).
Conclusion: Code generation is only appropriate for a relatively small number of problems, but when you're able to use one, it will save your sanity.
A code generator is useful if:
The cost of writing and maintaining the code generator is less than the cost of writing and maintaining the repetition that it is replacing.
The consistency gained by using a code generator will reduce errors to a degree that makes it worthwhile.
The extra problem of debugging generated code will not make debugging inefficient enough to outweigh the benefits from 1 and 2.
For domain-driven or multi-tier apps, code generation is a great way to create the initial model or data access layer. It can churn out the 250 entity classes in 30 seconds ( or in my case 750 classes in 5 minutes). This then leaves the programmer to focus on enhancing the model with relationships, business rules or deriving views within MVC.
The key thing here is when I say initial model. If you are relying on the code generation to maintain the code, then the real work is being done in the templates. (As stated by Max E.) And beware of that because there is risk and complexity in maintaining template-based code.
If you just want the data layer to be "automagically created" so you can "make the GUI work in 2 days", then I'd suggest going with a product/toolset which is geared towards the data-driven or two-tier application scenario.
Finally, keep in mind "garbage in=garbage out". If your entire data layer is homogeneous and does not abstract from the database, please please ask yourself why you are bothering to have a data layer at all. (Unless you need to look productive :) )
How 'bout an example of a good use of a code generator?
This uses t4 templates (a code generator built in to visual studio) to generate compressed css from .less files:
http://haacked.com/archive/2009/12/02/t4-template-for-less-css.aspx
Basically, it lets you define variables, real inheritance, and even behavior in your style sheets, and then create normal css from that at compile time.
Everyone talks here about simple code generation, but what about model-driven code generation (like MDSD or DSM)? This helps you move beyond the simple ORM/member accessors/boilerplate generators and into code generation of higher-level concepts for your problem domain.
It's not productive for one-off projects, but even for these, model-driven development introduces additional discipline, better understanding of employed solutions and usually a better evolution path.
Like 3GLs and OOP provided an increase in abstraction by generating large quantities of assembly code based on a higher level specification, model-driven development allows us to again increase the abstraction level, with yet another gain in productivity.
MetaEdit+ from MetaCase (mature) and ABSE from Isomeris (my project, in alpha, info at http://www.abse.info) are two technologies on the forefront of model-driven code generation.
What is needed really is a change in mindset (like OOP required in the 90's)...
I'm actually adding the finishing touches to a code generator I'm using for a project I've been hired on. We have a huge XML files of definitions and in a days worth of work I was able to generate over 500 C# classes. If I want to add functionality to all the classes, say I want to add an attribute to all the properties. I just add it to my code-gen, hit go, and bam! I'm done.
It's really nice, really.
There are many uses for code generation.
Writing code in a familiar language and generating code for a different target language.
GWT - Java -> Javascript
MonoTouch - C# -> Objective-C
Writing code at a higher level of abstraction.
Compilers
Domain Specific Languages
Automating repetitive tasks.
Data Access Layers
Initial Data Models
Ignoring all preconceived notions of code-generation, it is basically translating one representation (usually higher level) to another (usually lower level). Keeping that definition in mind, it is a very powerful tool to have.
The current state of programming languages has by no means reached its full potential and it never will. We will always be abstracting to get to a higher level than where we stand today. Code generation is what gets us there. We can either depend on the language creators to create that abstraction for us, or do it ourselves. Languages today are sophisticated enough to allow anybody to do it easily.
If with code generator you also intend snippets, try the difference between typing ctor + TAB and writing the constructor each time in your classes. Or check how much time you earn using the snippet to create a switch statement related to an enum with many values.
If you're paid by LOC and work for people who don't understand what code generation is, it makes a lot of sense. This is not a joke, by the way - I have worked with more than one programmer who employs this technique for exactly this purpose. Nobody gets paid by LOC formally any more (that I know of, anyway), but programmers are generally expected to be productive, and churning out large volumes of code can make someone look productive.
As an only slightly tangential point, I think this also explains the tendency of some coders to break a single logical unit of code into as many different classes as possible (ever inherit a project with LastName, FirstName and MiddleInitial classes?).
Here's some heresy:
If a task is so stupid that it can be automated at program writing time (i.e. source code can be generated by a script from, let's say XML) then the same can also be done at run-time (i.e. some representation of that XML can be interpreted at run-time) or using some meta-programming. So in essence, the programmer was lazy, did not attempt to solve the real problem but took the easy way out and wrote a code generator. In Java / C#, look at reflection, and in C++ look at templates
Related
Hi
We are going to start a CRUD project.
I have some experience using groovy and
I think it is the right tool.
My concern is about performance.
How good is groovy compared to a java solution.
It is estimated that we can have up to 100
simultaneosly users. We are going to use a
MySql DB and a tomcat server.
Any comment or suggestion?
Thanks
I've recently gathered five negative votes (!) on an answer on Groovy performance; however, I think there should be, indeed, a need for objective facts. Personally, I think it's productive and fun to work with Groovy and Grails; nevertheless, there is a performance issue that needs to be addressed.
There are a number of benchmark comparisons on the web, including this one. You can never trust single benchmarks (and the cited one isn't even close to being scientific), but you'll get the idea.
Groovy strongly relies on runtime meta programming. Every object in Groovy (well, except Groovy scripts) extends from GroovyObject with its invokeMethod(..) method, for example. Every time you call a method in your Groovy classes, the method will not be called, directly, as in Java, but by invoking the aforementioned invokeMethod(..) (which does a whole bunch of reflection and lookups).
Additionally, every GroovyObject has an associated MetaClass. The concepts of method invocation, etc., are similar.
There are other factors that decrease Groovy performance in comparison to Java, including boxing of primitive data types and (optional) weak typing, but the aforementioned concept of runtime meta programming is crucial. You cannot even think of a JIT compiler with Groovy, that compiles Java bytecode to native code to speed up execution.
To address these issues, there's the Groovy++ project. You simply annotate your Groovy classes with #Typed, and they'll be statically compiled to (real) Java bytecode. Unfortunately, however, I found Groovy++ to be not quite mature, and not well integrated with the main Groovy line, and IDEs. Groovy++ also contradicts basic Groovy programming paradigms. Moreover, Groovy++' #Typed annotation does not work recursively, that is, does not affect underlying libraries like GORM or the Grails controllers infrastructure.
I guess you're evaluating employing a Grails project, as well.
When looking at Grails' GORM, that framework makes heavily use of runtime meta programming, using Hibernate directly, should perform much better.
At the controllers or (especially) services level, extensive computations can be externalized to Java classes. However, GORMs proportion in typical CRUD applications is higher.
Potential performance in Grails are typically addressed by caching layers at the database level or by avoiding to call service or controllers methods (see the SpringCache plugin or the Cache Filter plugin). These are typically implemented on top of the Ehcache infrastructure.
Caching, obviously, may suit well with static data in contrast to (database) data that frequently changes, or web output that is rather variable.
And, finally, you can "throw hardware at it". :-)
In conclusion, the most decisive factor for or against using Groovy/Grails in a large-scaling website ought to be the question whether caching fits with the specific website's nature.
EDIT:
As for the question whether Java's JIT compiler had a chance to step in ...
A simple Groovy class
class Hello {
def getGreeting(name) {
"Hello " + name
}
}
gets compiled to
public class Hello
implements GroovyObject
{
public Hello()
{
Hello this;
CallSite[] arrayOfCallSite = $getCallSiteArray();
}
public Object getGreeting(Object name) {
CallSite[] arrayOfCallSite = $getCallSiteArray();
return arrayOfCallSite[0].call("Hello ", name);
}
static
{
Long tmp6_3 = Long.valueOf(0L);
__timeStamp__239_neverHappen1288962446391 = (Long)tmp6_3;
tmp6_3;
Long tmp20_17 = Long.valueOf(1288962446391L);
__timeStamp = (Long)tmp20_17;
tmp20_17;
return;
}
}
This is just the top of an iceberg. Jochen Theodoru, an active Groovy developer, put it that way:
A method invocation in Groovy consists
usually of several normal method
calls, where the arguments are stored
in a array, the classes of the
arguments must be retrieved, a key is
generated out of them, a hashmap is
used to lookup the method and if that
fails, then we have to test the
available methods for compatible
methods, select one of the methods
based on the runtime type, create a
key for the hasmap and then in the
end, do a reflection like call on the
method.
I really don't think that the JIT inlines such dynamic, highly complex invocations.
As for a "solution" to your question, there is no "do it that way and you're fine". Instead, the task is to identify the factors that are more crucial than others and possible alternatives and mitigation strategies, to evaluate their impact on your current use cases ("can I live with it?"), and, finally, to identify the mix of technologies that meets the requirements best (not completely).
Performance (in the context of web applications) is an aspect of your application and not of the framework/language you are using. Any discussion and comparison about method invocation speed, reflection speed and the amount of framework layers a call goes through is completely irrelevant. You are not implementing photoshop filters, fractals or a raytracer. You are implementing web based CRUD.
Your showstopper will most probably be inefficient database design, N+1 queries (in case you use ORM), full table scans etc.
To answer your question: use any modern language/web framework you feel more confident with and focus on correct architecture/design to solve the business problem at hand.
Thanks for the answers and advices. I like groovy. It might be some performance problems under some circumstances. Groovy++ might be a better choice. At his point I would prefer to give a chance to "spring roo" which has a huge overlapping with Groovy but you remain at java and NO roo.jar is added to your project. Therefore you are not paying any extra cost for using it.
Moreover "roo" allows backward engineering and roundtrip engineering.
Unfortunately the plug-in library is pretty small up to now.
Luis
50 to 100 active users is not much of a traffic. As long as you have cached pages correctly, mysql queries are properly indexes, you should be ok.
Here is a site I am running in my basement in a $1000 server. It's written in Grails.
Checkout performance yourself http://www.ewebhostguide.com
Caution: Sometimes Comcast connections are down and site may appear down. But that happens only for few minutes. Cons of running site in basement.
In theory, source code should not contain hardcoded values beyond 0, 1 and the empty string. In practice, I find it very hard to avoid all hardcoded values while being on very tight delivery times, so I end up using a few of them and feeling a little guilty.
How do you reconcile avoiding hardcoded values with tight delivery times?
To avoid hard-coding you should
use configuration files (put your values in XML or ini-like text files).
use database table(s) to store your values.
Of course not all values qualify to be moved to a config file. For those you should use constructs provided by the programming language such as (constants, enums, etc).
Just saw an answer to use "Constatn Interface". With all due respect to the poster and the voters, but this is not recommended. You can read more about that at:
http://en.wikipedia.org/wiki/Constant_interface
The assumption behind the question seems to me invalid.
For most software, configuration files are massively more difficult to change that source code. For widely installed, software, this could easily be a factor of a million times more difficult: there could easily be that many files hanging round on user installations which you have little knowledge and no control over.
Having numeric literals in the software is no different from having functional or algorithmic literals: it's just source code. It is the responsibility of any software that intends to be useful to get those values right.
Failing that make them at least maintainable: well named and organised.
Making them configurable is the kind of last-ditch compromise you might be forced into if you are on a tight schedule.
This comes with a little bit of planning, in most cases it is as simple as having a configuration file, or possibly a database table that stores critical configuration items. I don't find that there is any reason that you "have" to have hard coded values, and it shouldn't take you much additional time to offload to a configuration mechanism to where tight time lines would be a valid excuse.
The problem of hardcoded values is that sometimes it's not obvoius that particular code relies on them. For example, in java it is possible to move all constants into separate interface and separate particular constants into inner sub-interfaces. It's quite convenient and obvious. Also it's easy to find the constant usage just by using IDE facilities ("find usage" functionality) and change or refactor them.
Here's an example:
public interface IConstants {
public interface URL {
String ALL = "/**";
}
public interface REST_URL {
String DEBUG = "/debug";
String SYSTEM = "/system";
String GENERATE = "/generate";
}
}
Referencing is quite human readable: IConstants.REST_URL.SYSTEM
Most non-trivial enterprise-y projects will have some central concept of properties or configuration options, which already takes care of loading up option from a file/database. In these cases, it's usually simple (as in, less than 5 minutes' work) to extend this to support the new propert(ies) you need.
If your project doesn't have one, then either:
It could benefit from one - in which case write it, taking values from a flat .properties file to start with. This shouldn't take more than an hour, tops, and is reusable for any config stuff in future
Even that hour would be a waste - in which case, you can still hav a default value but allow this to be overridden by a system property. This require no infrastructure work and minimal time to implement in your class.
There's really no excuse for hardcoding values - it only saves you a few minutes at most, and if your project deadline is measured in minutes then you've got bigger problems than how to code for configurability.
Admittedly, I hardcode a lot of stuff at my current hobby project. Configuration files are ridiculously easy to use instead (at least with Python, which comes with a great and simple .cfg parser), I just don't bother to use them because I am 99% confident that I will never have to change them - and even if that assumption proved false, it's small enough to refactor it with reasonable effort. For annything larger/more important, however, I would never type if foo == "hardcoded bar", but rather if foo == cfg.bar (likely with a more meaningful name for cfg). Cfg is a global singleton (yeah, I know...) which is fed the .cfg file at startup, and next time some sentinel value changes, you change the configuration file and not the source.
With a dynamic/reflective language, you don't even need to change the part loading the .cfg when you add another value to it - make it populate the cfg object dynamically with all entries in the file (or use a hashmap, for that matter) and be done.
2 suggestions here:
First, if you are working on embedded system using language like C. Simply work out a coding convention to use a #define for any string or constant. All the #define should be categorized in a .h file. That should be enough - not too complex but enough for maintainability. You don't need to mangle all the constant between the code line.
Second, if you are working on a application with access to DB. It is simple just to keep all the constant values in the database. You just need a very simple interface API to do the retrieval.
With simple tricks, both methods can be extended to support multi-language feature.
Background: Eldridge asked me to explain what the difference is between the difference phases of time when it comes to writing and deploying code.
He wants to know:
1) what is the difference between: 1) design time; 2) compile time; 3) run-time?
2) what are specific examples of things a programmer would not be able to hard-wire into his code and not know until run-time?
3) are scripting languages with many run-time "tricks" better (other than personal preference reasons) for people who need more 'run-time' flexibility?
Question:
Although I have answers for Eldrige based on my own programming views, it seemed like a good idea to get different perspectives, so as not to give a "biased" answer. ... So, what unbiased answer can you give to explain these things (assuming whatever language[s]).
Okay, since no one else has tried, I'll take a crack at this.
Design time is the time spent creating the source files. Code in text files, form definitions, etc.
Compile time is the time spent in the various phases of compilation. Preprocessing, lexing/parsing, AST creation and optimization, code generation, and linking.
Run time is the time spent from when the executable is loaded until the memory used for the text pages is freed.
A programmer can hardwire anything they like, but it makes little sense for some things:
User's home directoy
Location of external database resources
Expected hardware configuration of the machine running the executable
etc.
Both static and dynamic languages (I personally hate the description "scripting languages") have a place in computing. There is nothing a dynamic language can do that a static language can't; the only difference is in the amount of code it takes to implement the functionality.
Design-time - I would say is when you use automated tools and graphic designers to generate your source code. Stuff like valid values for a control's property can be determined at 'design-time'
Compile-Time - Is things that can be determined by the compiler at compile-time, e.g. the compiler can determine that the following expression will always contain the constant value of '86400'.
const int SecsInDay = 60 * 60 * 24;
Run-Time - are for things that can only be determined while the program is running (e.g. by the VM). This includes any user input, dynamic configuration settings, etc.
Scripting languages are more flexible at creating mutating logic as it can by-pass compilation and create new behaviour based on input and logic at run-time, e.g. Java Script's eval() statement can execute any arbitrary javascript inputted by a user or returned via a webservice, etc.
I just started creating my data-access layer using LinqToSql. Everybody is talking about the cool syntax and I really like Linq in general.
But when I saw how your classes get generated if you drag some tables on the LinqContext I was amazed: So much code which nobody would need?!
So I looked how other people used LinqToSql, for example Rob Connery at his StoreFront Demo.
As I don't like the way that all this code is generated I created my domain layer by hand and used the generated classes as reference. With that solution I'm fine, as I can use the features provided by Linq (dereferred execution, lazy loading, ...) and my domain layer is quite easy to understand.
How are you using LinqToSql?
The created classes are not as heavy as it seems. Of course it takes quite a few lines of code, but all in all it is as lightweight as it can be for the features it's providing.
I used to create my own tables, too, but now instead I just use the LINQtoSQL DataContext. Why? Creating is simpler, the features are better, interoperability works, it is probably even faster than my own stuff (not in every aspect. Usually my own stuff was extremely fast in one thing, but the generic stuff was faster in everything else).
But the most important part: it is easier to bring new developers into the LINQ stuff than into my own. There are tutorials, example codes, documentation, everything, which I'd have to create for my code by myself. Same with using my stuff with other technology, like WCF or data binding. There are a lot of pitfalls to be taken care of.
I learned not to develop myself into a corner the hard way, it looks fast and easy at the start, is a lot more fun than learning how to use the libs, but is a real pain in the a after a few months down the road, usually even for myself.
After some time the novelty of creating my own data containers wears off, and I noticed the pain connected to adding a feature. A feature I would have had for free, if I had used the provided classes.
Next thing I had to explain my code to some other programmer. Had I used the provided classes, I could have pointed him to some web site to learn about the stuff. But for my classes, I had to train him himself, which took a long time and made it hard to get new people on a project.
LinqToSql generates a set of partial classes for your tables. You can add interface definitions to the 'other half' of these partial classes that implement your domain model.
Then, if you use the repository pattern to wrap access to the Linq queries, so that they return interface implementations of your objects (the underlying Linq objects), LinqToSql becomes quite flexible.
You can hand-write your own classes and either use the LINQ to SQL attributes to declare the mappings or an external XML file.
If you would like to stick with the existing designer and just modify the code generation process pick up my templates that let you tailor the generated code.
Use compiled queries. Linq to SQL is dog slow otherwise. Really.
We use our hand crafted domain model, along with the generated classes, plus a simple utility which utilizes reflection to convert between them when needed. We've contemplated writing a converter generator if we reach a point where reflection creates a performance bottleneck.
To take an example, consider a set of discounts available to a supermarket shopper.
We could define these rules as data in some standard fashion (lists of qualifying items, applicable dates, coupon codes) and write generic code to handle these. Or, we could write each as a chunk of code, which checks for the appropriate things given the customer's shopping list and returns any applicable discounts.
You could reasonably store the rules as objects, serialised into Blobs or stored in code files, so that each rule could choose its own division between data and code, to allow for future rules that wouldn't fit the type of generic processor considered above.
It's often easy to criticise code that mixes data in, via if statements that check for 6 different things that should be in a file or a database, but is there a rule that helps in the edge cases?
Or is this the point of Object Oriented design, to stop us worrying about the line between data and code?
To clarify, the underlying question is this: How would you code the above example? Is there a rule of thumb that made you decide what is data and what is code?
(Note: I know, code can be compiled, but in a world of dynamic languages and JIT compilation, even that is a blurry concept.)
Fundamentally, there is of course no difference between data and code, but for real software infrastructures, there can be a big difference. Apart from obvious things like, as you mentioned, compilation, the biggest issue is this:
Most sufficiently large projects are designed to produce "releases" that are one big bundle, produced in 3-month (or longer) cycles, tested extensively and cannot be changed afterwards except in tightly controlled ways. "Code" most definitely cannot be changed, so anything that does need to be changed has to be factored out and made "configuration data" so that changing it becomes palatable those whose job it is to ensure that a release works.
Of course, in most cases bad configuration data can break a release just as thoroughly as bad code, so the whole thing is largely an illusion - in reality it doesn't matter whether it's code or "configuration data" that changes, what matters is that the interface between the main system and the parts that change is narrow and well-defined enough to give you a good chance that the person who does the change understands all consequences of what he's doing.
This is already harder than most people think when it's really just a few strings and numbers that are configured (I've personally witnessed a production mainframe system crash because it had one boolean value set differently than another system it was talking to). When your "configuration data" contains complex logic, it's almost impossible to achieve. But the situation isn't going to be any better ust because you use a badly-designed ad hoc "rules configuration" language instead of "real" code.
This is a rather philosophical question (which I like) so I'll answer it in a philosophical way: with nothing much to back it up. ;)
Data is the part of a system that can change. Code defines behavior; the way in which data can change into new data.
To put it more accurately: Data can be described by two components: a description of what the datum is supposed to represent (for instance, a variable with a name and a type) and a value.
The value of the variable can change according to rules defined in code. The description does not change, of course, because if it does, we have a whole new piece of information.
The code itself does not change, unless requirements (what we expect of the system) change.
To a compiler (or a VM), code is actually the data on which it performs its operations. However, the to-be-compiled code does not specify behavior for the compiler, the compiler's own code does that.
It all depends on the requirement. If the data is like lookup data and changes frequently you dont really want to do it in code, but things like Day of the Week, should not chnage for the next 200 years or so, so code that.
You might consider changing your topic, as the first thing I thought of when I saw it, was the age old LISP discussion of code vs data. Lucky in Scheme code and data looks the same, but thats about it, you can never accidentally mix code with data as is very possible in LISP with unhygienic macros.
Data are information that are processed by instructions called Code. I'm not sure I feel there's a blurring in OOD, there are still properties (Data) and methods (Code). The OO theory encapsulates both into a gestalt entity called a Class but they are still discrete within the Class.
How flexible you want to make your code in a matter of choice. Including constant values (what you are doing by using if statements as described above) is inflexible without re-processing your source, whereas using dynamically sourced data is more flexible. Is either approach wrong? I would say it really depends on the circumstances. As Leppie said, there are certain 'data' points that are invariate, like the days of the week that can be hard coded but even there it may be advantageous to do it dynamically in certain circumstances.
In Lisp, your code is data, and your
data is code
In Prolog clauses are terms, and terms
are clauses.
The important note is that you want to separate out the part of your code that will execute the same every time, (i.e. applying a discount) from the part of your code which could change (i.e. the products to be discounted, or the % of the discount, etc.)
This is simply for safety. If a discount changes, you won't have to re-write your discount code, you'll only need to go into your discounts repository (DB, or app file, or xml file, or however you choose to implement it) and make a small change to a number.
Also, if the discount code is separated into an XML file, then you can give the entire application to a manager, and with sufficient instructions, they won't need to pester you whenever they want to change the discount rates.
When you mix in data and code, you are exponentially increasing the odds of breaking when anything changes. So, as leppie said, you need to extract the constantly changing parts, and put them in a separate place.
Huge difference. Data is a given to system while code is a part of system.
Wrong data is senseless: our code===handler is good and what you put that you take, it is not a trouble of system that you meant something else. But if code is bad - system is bad.
In example, let's consider some JSON, some bad code parser.js by me and let's say good V8. For my system bad parser.js is a code and my system works wrong. But for Google system my bad parser is data that no how says about quality of V8.
The question is very practical, no sophistic.
https://en.wikipedia.org/wiki/Systems_engineering tries to make good answer and money.
Data is information. It's not about where you decide to put it, be it a db, config file, config through code or inside the classes.
The same happens for behaviors / code. It's not about where you decide to put it or how you choose to represent it.
The line between data and code (program) is blurry. It's ultimately just a question of terminology - for example, you could say that data is everything that is not code. But, as you wrote, they can be happily mixed together (although usually it's better to keep them separate).
Code is any data which can be executed. Now since all data is used as input to some program at some point of time, it can be said that this data is executed by a program! Thus your program acts as a virtual machine for your data. Hence in theory there is no difference between data and code!
In the end what matters is software engineering/development considerations like performance, efficiency etc. For example data driven programs may not be as efficient as programs which have hard coded (and hence fragile) conditional statements. Hence I choose to define code as any data which can be efficiently executed and all else being plain data.
It's a tradeoff between flexibility and efficiency. Executable data (like XML rules) offers more flexibility (sometimes) while the same data/rules when coded as part of the application will run more efficiently but changing it frequently becomes cumbersome. In other words executable data is easy to deploy but is inefficient and vice-versa. So ultimately the decision rests with you - the software designer.
Please correct me if I wrong.
Relationship between code and data is as follows:
code after compiled to a program processes the data while execution
program can extract data, transform data, load data, generate data ...
Also
program can extract code, transform code, load code, generate code tooooooo...
Hence code without compiled or interperator is useless, data is always worth..., but code after compiled can do all the above activities....
For eg)
Sourcecontrolsystem process Sourcecodes
here source code itself is a code
Backupscripts process files
here files is a data and so on...
I would say that the distinction between data, code and configuration is something to be made within the context of a particular component. Sometimes it's obvious, sometimes less so.
For example, to a compiler, the source code it consumes and the object code it creates are both data - and should be separated from the compiler's own code.
In your case you seem to be describing the option of a particularly powerful configuration file, which can contain code. Much as, for example, the GIMP lets you 'configure' plugins using Scheme. As the developer of the component that reads this configuration, you would think of it as data. When working at a different level -- writing the configuration -- you would think of it as code.
This is a very powerful way of designing.
Applying this to the underlying question ("How would you code the above example?"), one option might be to adopt or design a high level Domain Specific Language (DSL) for specifying rules. At startup, or when first required, the server reads the rule and executes it.
Provide an admin interface allowing the administrator to
test a new rule file
replace the current configuration with that from a new rule file
... all of which would happen at runtime.
A DSL might be something as simple as a table parser or an XML parser, or it could be something as sophisticated as a scripting language. From C, it's easy to embed Python or Lua. From Java it's easy to embed Groovy or Clojure.
You could switch in compiled code at runtime, with clever linking or classloader tricks. This seems more difficult and less valuable than the embedded DSL option, in my opinion.
The best practical answer to this question I found is this:
Any class that needs to be serialized, now or in any foreseeable future, is data.
Everything else is code.
That's why, for example, Java's HashMap is data - although it has a lot of code, API methods and specific implementation (i.e., it might look as code at first glance).