PureBasic Forums - English

I'm accessing an XML database exported from a Java Properties file. The nodes are coded as key/value pairs thusly: I have extracted the pairs into a map, but it seems the values are stored as UTF-16 rather than the UTF-8 from the file. When I try to stuff a value into a Scintilla text box, it appears it wants the original UTF8.

Normally, this wouldn't be an issue. I can recode back to UTF8. However, the user can navigate through screens several times a second, depending on what they are doing, which makes for dozens of unnecessary allocations/deallocations and CPU cycles.

My questions are:
1. Are my worries about reencoding texts, which can go up to 64K and above in size, a tempest in a teapot?
2. Can I prevent the reencoding when extracting to the map?
3. If not, can I access the XML nodes directly by the attribute (key)? The nodes seem to be addressable only by number.
4. If I use a different XML schema, rather than abusing the attribute function like Java does here, would that be a better solution for querying the XML directly? E.g. This is all to avoid writing my own text gadget to have full access to keystrokes and mouse interaction, i.e., a custom context menu that integrates with spellcheck, dictionary, and other program features. This is the last problem to solve before I really dive in to coding an OOP program procedurally.

Better write your stuff the easy way and if you see any issue, find another solution (ie: optimize last). Re-encoding is ligthning fast and shouldn't be an issue.

There's a difference between premature optimization and just plain bad ideas. I'm trying to rule out the latter here. I can't do anything much about performance until I see the finished result and how it compares to the Java version.