[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: A complex case of profiling
On Wed, Apr 27, 2005 at 06:51:11PM +0200, Jirka Kosek <jirka@kosek.cz> wrote a message of 102 lines which said: > Can you be more specific about what is the problem? OK, here is the complete use case. I thought that people would prefer a "digest" version but here is the full one. I write the documentation for a software used by different domain names registries (like ".fr" or ".cz", although our users are typically african registries). Since every registry has a different registration policy, an important part of the documentation has to be specific to some options. The options are boolean (yes/no). Practical examples: some registries sell directly to the end user, some force to go through registrars. So, in the first case, the documentation will say: <para>We have three main entities in the database: Domains, Hosts and Contacts.</para> while in the second: <para>We have four main entities in the database: Domains, Hosts, Registrars and Contacts.</para> Also, some registries allow IDN (domain names in Unicode) and some do not. So, in the first case, we will have: <para>Legal domain names are those authorized by <rfc num="3490"/>, practically all the Unicode character repertoire.</para> while the second will be: <para>Legal domain names are composed of US-ASCII letters, digits and the hyphen character. Domains names starting with "xn--" are not allowed.</para> Now, how to implement it? Writing several documentations would be inconvenient, both because an important part is common to all options and also because of the combinatorial explosion (most combinations of options are valid). I thought to use the "condition" common attribute, something like: <para>Domain names have syntactic restrictions. For instance, spaces are never authorized. <phrase condition="idn">Legal domain names are those authorized by <rfc num="3490"/>, practically all the Unicode character repertoire.</phrase> <phrase condition="no_idn">Legal domain names are composed of US-ASCII letters, digits and the hyphen character because of <rfc num="1123"/>. Domains names starting with "xn--" are not allowed. </phrase> </para> But the profiling system, as documented in http://www.sagehill.net/docbookxsl/Profiling.html does not work because: 1) it does not allow negated options like no_idn above. 2) it does not allow to combine options (say a sentence mus appear only if "registrars" and "idn") although I may simply use nested <phrases> with conditions. So, I planned to write a DOM program which will process the elements with "condition" to my liking but I wonder if there is a better way.
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]