Talk:XML/Output

You need to separate the task description fully from the example so others can follow the description alone. --Paddy3118 19:43, 28 December 2008 (UTC)

How about extending the task by adding a 'special' name that needs escaping, such as '<None>', and ask for sample output. You would then force entries to handle escaped characters. --Paddy3118 05:15, 8 June 2009 (UTC)

On one level that'd be good, but on another it's not so good. (XML docs aren't supposed to contain magic flag values in text nodes; that's what using a different elements is for...) —Donal Fellows 09:00, 8 June 2009 (UTC)

Why not simply have a name in the list which includes something with a named entity, like the German "Jürgen" ("J&uuml;rgen") or the french "André" ("Andr&eacute;")? --Ce 09:18, 8 June 2009 (UTC)

Because those don't need to be encoded; the data stream might be UTF-8 (which is actually the default for XML...) For the quoting to work, we'd need to force the inclusion of each of <>'"& (those are the only ones that XML requires handling, and even the quoting characters only need doing in some circumstances; it's really messy) and those characters don't look like first names. (The set of things you're supposed to quote is larger in HTML.) —Donal Fellows 10:47, 8 June 2009 (UTC)

Of course, the task description contains: "If attempting this task by direct string manipulation, the implementation must include code to perform entity substitution for the characters that have entities defined in the XML 1.0 specification." That is, the task says, encode whereever you can, not wherever you must. --Ce 11:54, 8 June 2009 (UTC)

What is the point?

If the task requires creating some fixed output, the easiest and most effective way would be just to insert the required contents in the file. That has very little to do with real world problems.

It would make more sense if the task would be for example to read list of names from a text file and use it to create an XML file with specified structure. In that case it would make sense to to require converting special characters into entities, since such special characters might exist in the input file. In order to compare the output, you could then have an example input file. --PauliKL 11:48, 8 June 2009 (UTC)

Instead of reading from a file, just get it from standard input (or more generally, "from the user"). That way, the task doesn't get cluttered with file system stuff unrelated to this task. --Ce 11:54, 8 June 2009 (UTC)

Modification of Task

I suggest the task be changed to adress some of the issues discussed above. I would suggest the following

---

Create a function that takes a list of character names and a list of corresponding remarks and returns an XML doccument of <Character> elements each with a name attributes and surrounding their remarks. All <Character> elements are to be enclosed in an outer <CharacterQuotes> element.

As an example, calling the function with names of: “April”, “Tam O'Shanter”, and “Emily” and remarks of:

“Bubbly: I\'m > Tam and <= Emily”
“Burns: "When chapman billies leave the street ..."”
and “Short & shrift”

Should produce the XML (modulo line breaks):

<CharacterQuotes>
    <Character name="April">Bubbly: I'm &gt; Tam and &lt;= Emily</Character>
    <Character name="Tam O&apos;Shanter"
            >Burns: &quot;When chapman billies leave the street ...&quot;</Character>
    <Character name="Emily">Short &amp; shrift</Character>
</CharacterQuotes>

Note: the example is chosen to also show correct escaping of XML strings

--Paddy3118 07:14, 9 June 2009 (UTC)

The suggested task description is not quite clear to a non-English speaking person like me. Maybe splitting it into shorter sentences would help. Anyway, I think I got the idea from the examples. But I think the example input should not contain language specific details such as the \ escape character. And not the quotes (string delimiters) either, if they are not supposed to be part of the input text. --PauliKL 09:42, 9 June 2009 (UTC)

Hi Paul, I'll work on the description, but on the use of \: I couldn't think of a way to include a quote in a quoted string without it - I am open to sugestions though. --Paddy3118 11:11, 9 June 2009 (UTC)

That's the point when I use UNICODE quotes. Thankfully they're easy to type on this keyboard. —Donal Fellows 12:26, 9 June 2009 (UTC)

Why quoted strings? Why not just enter the example input as plain text? In real world applications, the data would came from user, or from file, or from stdin, etc. Normally, the text is just plain text, with no quotes or escape characters. (Of course the test code that calls the actual function could use string constants, but that is language and implementation specific.) On the other hand, using two input lists would be quite strange way to enter data in real world. It would be more logical to enter the name and remark together, as a pair. (But I don't quite understand what this "remark" is supposed to be.) --PauliKL 15:27, 9 June 2009 (UTC)