At what point does a config file become a programming language?

问题

I have been mulling over config files and their relationship to code for a while now and depending on the day and direction of the wind my opinions seem to change. More and more though I keep coming back to the realization I first had while learning Lisp: there is little difference between data and code. This seems doubly true for config files. When looked at in the right light a Perl script is little more than a config file for perl. This tends to have fairly heavy consequences for tasks such as QA and divisions of labor like who should be responsible for changing config files.

The creep from config file to full fledged language is generally slow and seems to be driven by the desire to have a generic system. Most projects seem to start out small with a few config items like where to write logs, where to look for data, user names and passwords, etc. But then they start to grow: features start to be able to be turned on or off, the timings and order of operations start to be controlled, and, inevitably, someone wants to start adding logic to it (e.g. use 10 if the machine is X and 15 if the machine is Y). At a certain point the config file becomes a domain specific language, and a poorly written one at that.

Now that I have rambled on to set the stage, here are my questions:

1. What is the true purpose of a config file?
2. Should an attempt be made to keep config files simple?
3. Who should be responsible for making changes to them (developers, users, admins, etc.)?
4. Should they be source controlled (see question 3)?

As I said earlier my answers to these questions shift constantly, but right now I am thinking:

1. to allow a non-programmers to change large chunks of behaviour quickly
2. yes, anything that is not coarsely grained should be in code
3. users should be responsible for config files and programmers should be responsible for a configuration layer between config files and code that gives more fine grained control of the application
4. no, but the finer grained middle layer should be

回答1:

Very interesting questions!

I tend to limit my config files to a very simple "key=value" format, because I fully agree with you that config files can very quickly become full-blown programs. For example, anyone who has ever tried to "configure" OpenSER knows the feeling you are talking about: it's not configuration, it's (painful) programming.

When you need your application to be very "configurable" in ways that you cannot imagine today, then what you really need is a plugins system. You need to develop your application in a way that someone else can code a new plugin and hook it into your application in the future.

So, to answer your questions:

1. What is the true purpose of a config file?

   I would say, to allow the people who will install your application to be able to tweek some deployment-related parameters, such as host name, number of threads, names of the plugins you need, and the deployment-parameters for those plugins (check out FreeRadius's configuration for an example of this principle), etc.. Definitely not the place to express business logic.

2. Should an attempt be made to keep config files simple?

   Definitely. As you suggested, "programming" in a config file is horrible. I believe it should be avoided.

3. Who should be responsible for making changes to them (developers, users, admins, etc.)?

   In general, I would say admins, who deploy the application.

4. Should they be source controlled (see question 3)?

   I usually don't source-control the configuration files themselves, but I do source-control a template configuration file, with all the parameters and their default values, and comments describing what they do. For example, if a configuration file is named database.conf, I usually source-control a file named database.conf.template. Now of course I am talking about what I do as a developer. As an admin, I may want to source-control the actual settings that I chose for each installation. For example, we manage a few hundred servers remotely, and we need to keep track of their configurations: we chose to do this with source-control.

---

Edit: Although I believe the above to be true for most applications, there are always exceptions, of course. Your application may allow its users to dynamically configure complex rules, for example. Most email clients allow the users to define rules for the management of their emails (for example, "all emails coming from 'john doe' and not having me in the To: field should be discarded"). Another example is an application that allows the user to define a new complex commercial offer. You may also think about applications like Cognos which allow their users to build complex database reports. The email client will probably offer the user a simple interface to define the rules, and this will generate a complex configuration file (or even perhaps a bit of code). On the other hand, the user-defined configuration for the commercial offers might be saved in a database, in a structured way (neither a simple key=value structure nor a portion of code). And some other applications might even allow the user to code in python or VB, or some other automation-capable language. In other words... your mileage may vary.

回答2:

Ok. You will have some users which want a really simple config, you should give it to them. At the same time, you will have constant requests of "Can you add this? How do I do in the config file?", I don't see why you can't support both groups.

The project I am currently working on uses Lua for its configuration file. Lua is a scripting language, and it works quite well in this scenario. There is available an example of our default configuration.

You'll note that it is mainly key=value statements, where value can be any of Lua's built-in types. The most complicated thing there are lists, and they aren't really complicated (it's just a matter of syntax).

Now I'm just waiting for someone to ask how to set their server's port to a random value every time they start it up...

回答3:

Recently I was working upon a project and I realised that I wanted to have conditionals inside my configuration file - which had previously just been a pretty simple one of the form:

key = val
key2 = val
name = `hostname`

I didn't want to write a mini-language, because unless I did it very carefully I couldn't allow the flexibility that would be useful.

Instead I decided that I'd have two forms:

1. If the file started with "#!" and was executable I'd parse the result of running it.

2. Otherwise I'd read it as-is

This means that I can now allow people to write "configuration files" that look like this:

 #!/usr/bin/perl
if ( -x /bin/foo ) 
{
   print <<EOF;
foo=me
bar=you
EOF
}
else
{
   print <<EOF;
foo=bar
bar=foo
EOF
}

This way I get the power of a dynamic configuration file if the user wants to use it, and the simplicity of not having to write my own mini-language.

回答4:

Every (sufficiently-long-lived) config file schema eventually becomes a programming language. Due to all the implications you describe, it is wise for the config-file designer to realize she is authoring a programming language and plan accordingly, lest she burden future users with bad legacy.

回答5:

I have a different philosophy about config files. Data about how an application should be run is still data, and therefore belongs in a data store, not in code (a config file IMO is code). If end users need to be able to change the data, then the application should provide an interface to do so.

I only use config files to point at data stores.

回答6:

You could turn to theory of computation to define what counts as a programming language. If your configuration file format is Turing Complete then it reasonably counts as a programming language. By this definition, a file format to describe levels of Sokoban counts as a programming language (see here). There are other levels of complexity below Turing Complete that may also count, such as Regular Grammars and Pushdown Automata.

Another way to look at it is that many config files are only capable of data markup, whereas a proper programming language must be able to implement algorithms. For example, JSON is a config file format, whereas ECMA Script is a programming language.

回答7:

Here's my thoughts:

1. To allow the runtime behavior of an application to be modified easily. This can be by programmers or non programmers, depending on the needs. This can be during development, but I often view configuration files as a way to help make a program more flexible at any point.

2. Yes. I think config files should be as simple as possible, given the constraint that you may need various options to control different behaviors of your runtime. I prefer grouping configuration settings, and simplifying them as much as possible.

3. Depends on what and why the change is being made. If users are going to be changing it, a front-end should be made to hide them from the details. The same is often true of non-developers in general.

4. I often source control the "default" configuration, but have a way to override this per system for the actual runtime.

As for adding logic to the config file - I'd avoid this. I think it's better to just have the configuration file switch on the logic in your application. Behavior in config files leads to a lack of maintainability and understanding, in my experience. I strongly prefer keeping configuration files as simple as possible.

回答8:

I tend to agree with the premise of this question. I avoid getting myself into trouble by predicting early that this is going to happen, and therefore never roll my own config system.

• Either I use the operating systems' config facuility (such as a plist, or gconf or whatever is appropriate),
• Or a simple flat file, as can be handled by something like an off the shelf INI parser.
• Bite the bullet and plug a light weight language parser, usually lua, sometimes tcl into the application,
• Or store data in a SQLite or similar relational database.

And resign myself to live with whatever decision I made, or if i cant, refactor to use one of the above choices that better suits the application.

Point is, there's not really any reason to use a home-grown config solution. For one thing, it's harder on your users to have to learn a new, application specific config format. For another, You benefit from all the many bug-fixes and updates that come free when using an off-the-shelf solution. Finally, Feature creep is put to rest, because, well, you actually can't just add one more feature without really doing a major overhaul cause the config system isn't really in your hands in the first place.

回答9:

It depends on what you agree with other developers on the team. Are you using config files just as config files or you are creating a Model Driven application.

Symptoms of config file becoming a programming language:

• name=value pairs start to depend on each other
• you feel a need to have flow control (ex. if (this) than that)
• documentation for config file becomes essential in order to do further development (instead of just using the application)
• before value from config is read it requires to have some context (i.e. values depend on something external to config file itself)

回答10:

Config files invariably inch their way to becoming ugly, illogical "full fledged programming languages". It takes art and skill to design good programming languages, and config languages turned programming language tend to be horrendous.

A good approach is to use a nicely designed language, say python or ruby, and use it to create a DSL for your configuration. That way your configuration language can remain simple on the surface but in actually be the full fledged programming language.

回答11:

I believe your question is very relevant given the move to "fluent interfaces". Many developers have "seen the light" with respect to XML configured applications. Using XML can be very verbose and difficult to edit correctly (especially if no schema is provided). Having a fluent interface allows the developer to configure the application in a domain-specific language with the assistance of some key-value pairs from a plain text configuration file (or perhaps command-line parameters). It also makes it very easy to setup and configure new instances of the application for testing or whatever.

Here are my answers to your question:

• What is the true purpose of a config file?

A config file is a way to allow the user to customize the behavior of their program at run-time.

• Should an attempt be made to keep config files simple?

Ideally, I would think that config files should at least be supplemented by a fluent interface to configure the program (this is useful in many respects). If you do require a config file then it should be kept very simple, nothing other than key-value pairs.

• Who should be responsible for making changes to them (developers, users, admins, etc.)?

I think the answer to this depends on your organization. It should be the responsibility of the person deploying the software to ensure that it is properly configured.

• Should they be source controlled (see question 3)?

I will steal this answer from someone else :) I like the idea of storing a template configuration in source control and modifying it for each local user's needs. Chances are one developer's config file is another developer's nightmare so it is best to leave things that vary by user out of source control. Having a template is also a nice way to let the person deploying the application (or other developers) see exactly what values are valid for the config file.

回答12:

I have seen python programs where the config file is code. If you don't need to do anything special (conditionals, etc.) it doesn't look much different from other config styles. e.g. I could make a file config.py with stuff like:

num_threads = 13
hostname = 'myhost'

and the only burden on the user, compared with (say) INI files, is that they need to put '' around strings. No doubt you could do the same thing in other interpreted languages. It gives you unlimited ability to complicate your config file if necessary, at the risk of possibly scaring your users.

回答13:

Yes, config files should be simple. They should contain no 'logic' themselves - think of them as a list of expressions in if statements, not the conditional statements in their entirety.

They're there to allow the user to decide which of the options coded within the application should be used, so don't try to make them complicated, it'll end up being self-defeating - you may end up writing simple config files to control how the original config file should be configured otherwise!

回答14:

One of the purposes of the "Oslo" work at Microsoft is to permit (though not require) resolution of this issue.

1. An application would ship with models of any new components it includes. It would also use existing models. For instance, it might include a web service, so it could reuse the system model of a web service.
2. The models will include metadata describing them, including enough information for tools to access them, either textually or graphically.
3. Parts of the models will correspond to "configuration"

This means that the equivalent of todays configuration files may be rich enough to support both textual and graphical editing of their configuration. The graphical tool will be supplied with "Oslo" (code name "Quadrant").

回答15:

I'll be the contrarian and submit it's only a language when it embodies more than can be represented by XML; or else when XML is considered to be a language.

Alternatively, most config files can be thought of as classes, but with only properties and no methods. And without methods, I don't think it's a language.

Ultimately, "language" is a squishy abstraction, but yes, the edges are ambiguous.

回答16:

The code of our applications becomes less important... There is scripting, there are all kind of attributes that define the behaviour of classes, methods, method arguments and properties. Users can define database triggers and database constraints. There can be very complicated config files. Sometimes the user can define XSLT stylsheets to manipulate input and output because our systems need to be open (SOA). And there is stuff like BizzTalk that needs complex configuration too. Users can define complex workflows.

We have to write better code to deal with this complex environment, so the code of our applications becomes more important...

回答17:

I'm a big fan of using python programs as config files, especially for daemons. I like to take the tack of making the daemon completely empty of configuration except for the "configuration port". The python program then connects to the daemon and proceeds to create objects in the daemon and wire them together to create the desired configuration. Once everything is set up, the daemon can then be left to run on it's own. The benefits, of course, are that you get a full fledged programming language to write your config files and since you already have a way to talk to the daemon from another program, you can use it for debugging and getting stats. The major downside is having to deal with messages from another program coming in at any time.

回答18:

Config file: "What is my purpose?"
You: "Configure the butter."
Config file: "Ok..."
Config file: "What is my purpose?"
You: "You configure butter."
Config file: "Oh my god."

1. There is no "true purpose" of a configuration file. Its whatever makes sense for your application. In general, things that differ (or might differ) between machines and don't change in the middle of your application run should probably be in a configuration file. Defaults, ports, and addresses for other services are all great candidates. Keys and secrets are also great candidates but should be handled separately from your normal config for security reasons. I disagree that the purpose of a config file is to allow quick changes to be made. The purpose should be to allow flexibility in the setup of your application. If a config file is a quick easy to way to allow that flexibility, so much the better - but you should not be intending your config files to be frequently changing.

2. Yes and no. Should you atempt to make your application's code simple? Yes. You should attempt to make everything you write simple and to the point. No more complicated than it needs to be. Same is true of your config. However, this is very application specific. Hardcoding what should be in config because it would make your config "too complicated" is bad design. In fact, trying to "keep things simple" is why config files end up being a giant mess. Sometimes the simplest move is to modularize. This is why your configuration files should be written in a well known general purpose programming langauge - not some terrible configuration language (read: all "configuration languages" suck).

3. Again, who should be modifying config files is completely application dependent. But I agree with miniquark, whoever is deploying the application should be in charge of the configuration.

4. Source control everything you can. Source control is great. You can roll stuff back super easily and you have a full history of the changes you've made and a record of who made those changes. So why not?

来源：https://stackoverflow.com/questions/648246/at-what-point-does-a-config-file-become-a-programming-language