|
List Info
Thread: Standard configuration files
|
|
| Standard configuration files |
|
2006-06-12 19:53:44 |
|
Hello Patrick, First, sorry for the off-topic, but you didn't tell me which mailing list is the right one. I reviewed all of them and from the topic of the list: "This mailing list is for discussing configuration tool development." I thought it was the right one.
Second, I agree that the task of standarizing the way any software get's its configuration can be a huge task. The point I'm trying to make is the lack of guidelines, much less standards of configuring services from their configuration files. Wouldn't it be great if there was a tool developers could use to configure their programs using standard, human readable, easy to understand configuration files? Most services only need a plain, name=value type file so there's no reason why most of them are different.
Doing a quick review of the configuration files for very common services this is what we have:
Service Type
Samba name=value with sections [].
Postfix name=value.
Apache name value with sections <>.
Amd name=value with sections [].
Dovecot name=value with sections {}, and subections.
Grub ; name=value with sections name value.
Spamassassin name value.
But even for the same service MTA for example, the configuration is completely different from one another: sendmail, postfix, qmail.
You guess is right, I don't have "a whole lot of experience developing configuration systems in software applications", I would say I have some experience. Apart from being a web developer (mainly), I have experience in systems administration, specially *nix. Although I think this is not relevant.
The XML-DTD standard was created to be able to exchange ALL kinds of data, and it does it very well... except for non-structured binary data. I'm sure it can be used to describe most, if not all services' configuration.
The question I'd like to ask you, the experts, is: do you think it is worth? I think it is. At least there should be a guide, or recomendations on this.
Thanks again for your attention. Best regards,
Pedro.
Date: Sat, 10 Jun 2006 22:28:19 -0500
From: "Patrick W. Barnes" <n-man.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">nman64 n-man.com>
Subject: Re: Standard configuration files
To: redhat.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
fedora-config-listredhat.com
Cc: Pedro Bezunartea L?pez <bezunartea.net" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">pedrobezunartea.net>
On Thursday 08 June 2006 19:02, "Pedro Bezunartea López";
<bezunartea.net" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
pedrobezunartea.net> wrote:
>; Hi!
>
> I've been working with Linux for a few years now and one of the things I
> like most about it is the possibility of configuring everything by
> modifying readable text files. This is good but if you need to configure
> several services you may ask yoursel: why are all the configuration files
> for all the services so different? couldn't we come out with a standard
> file format that all services, devices, etc would use? What do you think?
> Here some ideas to begin with:
> 1. It'd have to be based in XML-dtd. Two important reasons, first it's a
> widely used standard, and second it's readable enough... and I can think of
> another one: it can describe very complex models.
> 2. There are already some services that use this kind of configuration
> files: tomcat... there's a place to start.
>;
> Any suggestions are welcomed,
>
First, this thread is off-topic for this list:
http://fedoraproject.org/wiki/PostIsOffTopic
I'm going to guess you don't have a whole lot of experience developing
configuration systems in software applications. When developing an
application, a lot of consideration must go into the configuration system to
be used. ; The developer will generally make a selection based upon their
needs, knowledge and available tools. Sometimes, an XML solution might be
the wise choice, but there are also cases where nothing more than a flat-file
with "name=value"; pairs is necessary, cases where a more powerful database is
needed, and cases where some other structured format or scripting language
might be needed. The number of factors that must be considered is
astronomical. ; Making a choice for the wrong reasons can have catastrophic
consequences. There is no one right way, and there won't be in the
foreseeable future.
--
Patrick "The N-Man"; Barnes
n-man.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">nman64n-man.com
http://www.n-man.com/
LinkedIn:
http://www.linkedin.com/in/nman64
Have I been helpful? Rate my assistance!
http://rate.affero.net/nman64/
--
|
| Standard configuration files |
|
2006-06-15 16:41:57 |
|
Pedro,
Don't let this stuff get you down! ; I am learning just like you. The only ignorant question is one that left unasked. Ignorance can be fixed while stupid is forever. Don't stop making suggestiions either! Work with your idea and see if you really have something there. There are multiple paths to almost all goals. Do what works for you! That's what Linux is all about anyway each person helping the other and all of us growing in the process. Besides there are several flavors of Linux for those with discriminating tastes!
God bless,
Al
On 6/12/06, Pedro Bezunartea López <bezunartea.net">pedrobezunartea.net> wrote:
Hello Patrick, First, sorry for the off-topic, but you didn't tell me which mailing list is the right one. I reviewed all of them and from the topic of the list: "This mailing list is for discussing configuration tool development." I thought it was the right one.
Second, I agree that the task of standarizing the way any software get's its configuration can be a huge task. The point I'm trying to make is the lack of guidelines, much less standards of configuring services from their configuration files. Wouldn't it be great if there was a tool developers could use to configure their programs using standard, human readable, easy to understand configuration files? Most services only need a plain, name=value type file so there's no reason why most of them are different.
Doing a quick review of the configuration files for very common services this is what we have:
Service Type
Samba name=value with sections [].
Postfix name=value.
Apache name value with sections <>.
Amd name=value with sections [].
Dovecot name=value with sections {}, and subections.
Grub ; name=value with sections name value.
Spamassassin name value.
But even for the same service MTA for example, the configuration is completely different from one another: sendmail, postfix, qmail.
You guess is right, I don't have "a whole lot of experience developing configuration systems in software applications", I would say I have some experience. Apart from being a web developer (mainly), I have experience in systems administration, specially *nix. Although I think this is not relevant.
The XML-DTD standard was created to be able to exchange ALL kinds of data, and it does it very well... except for non-structured binary data. I'm sure it can be used to describe most, if not all services' configuration.
The question I'd like to ask you, the experts, is: do you think it is worth? I think it is. At least there should be a guide, or recomendations on this.
Thanks again for your attention. Best regards,
Pedro.
On Thursday 08 June 2006 19:02, "Pedro Bezunartea López";
<bezunartea.net" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
pedrobezunartea.net> wrote:
>; Hi!
>
> I've been working with Linux for a few years now and one of the things I
> like most about it is the possibility of configuring everything by
> modifying readable text files. This is good but if you need to configure
> several services you may ask yoursel: why are all the configuration files
> for all the services so different? couldn't we come out with a standard
> file format that all services, devices, etc would use? What do you think?
> Here some ideas to begin with:
> 1. It'd have to be based in XML-dtd. Two important reasons, first it's a
> widely used standard, and second it's readable enough... and I can think of
> another one: it can describe very complex models.
> 2. There are already some services that use this kind of configuration
> files: tomcat... there's a place to start.
>;
> Any suggestions are welcomed,
>
First, this thread is off-topic for this list:
http://fedoraproject.org/wiki/PostIsOffTopic
I'm going to guess you don't have a whole lot of experience developing
configuration systems in software applications. When developing an
application, a lot of consideration must go into the configuration system to
be used. ; The developer will generally make a selection based upon their
needs, knowledge and available tools. Sometimes, an XML solution might be
the wise choice, but there are also cases where nothing more than a flat-file
with "name=value"; pairs is necessary, cases where a more powerful database is
needed, and cases where some other structured format or scripting language
might be needed. The number of factors that must be considered is
astronomical. ; Making a choice for the wrong reasons can have catastrophic
consequences. There is no one right way, and there won't be in the
foreseeable future.
--
Patrick "The N-Man"; Barnes
n-man.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">nman64n-man.com
http://www.n-man.com/
LinkedIn:
http://www.linkedin.com/in/nman64
Have I been helpful? Rate my assistance!
http://rate.affero.net/nman64/
--
--
Fedora-config-list mailing list
redhat.com">Fedora-config-listredhat.com
https://www.redhat.com/mailman/listinfo/fedora-config-list
|
[1-2]
|
|