Text nodes limited to 4096 characters in length.

[openlayers]

When parsing XML, we have to be aware of a browser limitation. Given XML that has a text node longer than 4096 characters, the text will be split among multiple child nodes of the original parent. For parsing long linestrings (and linearrings) this poses a problem.

[crschmidt]

Please post both the SVG and, if possible, the GML.

[openlayers]

SVG generated by 2.4

[openlayers]

SVG generated by 2.5rc1

[openlayers]

Actually, further investigation reveals that the SVG contains "NaN" as the last coordinate, as generated by 2.5rc1.

[openlayers]

GML used as source for SVG layer

[tschaub]

The browser is imposing a 4kb (4096 character limit) on child nodes. We can work around this, but I have a hard time believing it is a regression (that it would have worked in 2.4). If someone can demonstrate that the old GML parser handled this, I'll get on it for 2.5 - but otherwise, this may slip to 2.6.

[openlayers]

If that's the case, then it's probably the extra digits of precision that appeared in 2.5 that are taking it over the limit. I'll look at it more during daylight hours.

(pete@…)

[tschaub]

The fix doesn't look that bad. I'll post something if I get it together. Are you sure the extra digits didn't appear on your end? It's the length of the GML node that is a problem - before parsing.

[tschaub]

concatenate value of all child nodes

[tschaub]

The node value patch fixes the problem. Firefox (and I assume others) truncates a text node to 4kb, spreading the text among multiple children. So where we had a single node before, we now have many. Again, this doesn't seem to me to be a regression.

The attach patch gives the XML parser a concatChildValues method. This returns the concatenated value of all child nodes - giving you what you'd expect for text nodes over 4096 characters in length. No tests for me tonight, but I'm marking review so that others can weigh in.

[crschmidt]

I don't have a problem with it, but we should definitely have a test before we put it in. It seems like the right way to go.

[openlayers]

I can confirm that the patch solved the problem here on my end. (pete@…)

[crschmidt]

I put together a test for this, but decided that sticking 4k of data in a test file was not the right way to go about this. I think we should rethink our methods of testing the various format classes: with the manual testing we've done, I'm happy to have it go in, and we just need to be aware that we need tests for it -- NeedsTests lists it so we don't forget.

Pete, thanks for the report and the feedback.

[tschaub]

All existing tests pass in IE and FF.

[tschaub]

(In [4410]) Adding a concatChildValues method to the XML parser. This gets around a messy 4kb limit for text node lengths - over which browsers split values among multiple siblings (see #1006).

[crschmidt]

(In [4432]) Pullup changes for RC3:

• KML Examples broken (Closes #1001)
• drag handler doesn't reset properties on left or improperly modified mousedown (Closes #1003)
• Text nodes limited to 4096 in length (Closes #1006)
• KML/GML Attribute CDATA Parsing (Closes #1007)
• Control.MousePosition (Closes #1008)