1

I am working on a project where there are a whole bunch of strings being stored in an XML file for later retrieval. Stored with each string are between 0-10 tags that a user can use to access and sort these strings.

The general use case that I am working towards is the user entering a "to find" tag and the app using the XMLREADER functionality to run through and identify all those strings with matching tags.

My Question: as I design the XML Schema, are there any known performance differences between reading an Element's name vs its attribute? For example if I design my Schema like this:

<requirement>
  <string>Here is the string text</string>
  <tags>
    <tag id='technology' />
    <tag id='users' />
  </tags>
</requirement>

versus

<requirement>
  <string>Here is the string text</string>
  <tags>
    <technology />
    <users />
  </tags>
</requirement>

And I am scanning through looking for all matching "technology" tags. Based on how the XMLREADER class works in C# is one going to give me a significant performance benefit over the other (all other variables being equal)?

Improve this question

2 Answers 2

Reset to default
5

In the first case, you'll have to either create a lookup or to loop through the tags to find a specific tag. In the second case, .NET Framework does that for you when processing the XML file.

For instance, if you want to find multiple times (and by multiple, I mean a lot) whether a requirement has a specific tag, the second case usage will be straightforward; the first one will require additional work of creating the lookup in the first place.

Of course, the difference in performance will start to be noticeable when you work with hundreds of thousands of requirements, containing each hundreds of thousands of tags. At this scale, you'll be able to implement both solutions, do the benchmarks and determine which one is faster, eventually profiling both and removing the bottlenecks before choosing the one you want to keep.

Until then, what you are doing is premature optimization. Based on speculative assertions, you'll probably end up writing more code and making it difficult to read and maintain, in the name of a few microseconds of runtime eventually saved (but never actually measured).

The reasons to chose the XML structure is usually based on how well this structure represents the data. In your case, both are quite representative, since in both cases, we understand well that a requirement contains tags technology and users. As for the tools, most will be able to handle both equally well. Think about using XSLT to get a list of tags, or CSS to style tags on a page: you can do it in both cases easily.

One benefit of the first approach may be when used with statically-typed languages. But even then, most frameworks will let you manipulate the second structure without having to use Reflection.

Improve this answer
1
  • Thanks @MainMa, the second is obviously easier to program as you have said, I just haven't personally seen many XML Schemas that used just the Name without any attributes are child elements. Was just wondering if there was some hidden reason why! Commented Apr 4, 2015 at 19:53
0

To improve performance and simulate multiple nodes with regardless of its XML size, you can use XSLT with XML and then use xpath to retrieve nodes. This would be much better.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.