This changes how we handle defaults so that we can parse an entire XML node normally before needing to use any of its attributes to determine what default values that node should have.
This comes with a slight performance impact:
~668ms to run parsexml in the old system on a large marker pack
~680ms to run parsexml in the new system on the same large marker pack
The slowdown is likely do to a combination of:
Creating a new Category object for each node
Initializing all the boolean values in the new Category object
Iterating over every attribute when doing an overlay
But the tradeoffs are worth it, especially because those performance hits can probably be optimized away in the future.
This overlay/underlay system will allow us to get simpler fine grain control over each field. We no longer need to do a search for a particular field, such as category's "name" field, we can just parse the whole category node using the normal attribute parsing tech and then access the value from a member variable. EG: We were previously looking for just the "name" attribute, but now we can look for any xml tag that we have associated with the 'name' field, such as 'bh-name'. All without adding custom logic to the parser.
This changes how we handle defaults so that we can parse an entire XML node normally before needing to use any of its attributes to determine what default values that node should have.
This comes with a slight performance impact:
~668ms to run parsexml in the old system on a large marker pack ~680ms to run parsexml in the new system on the same large marker pack
The slowdown is likely do to a combination of:
Categoryobject for each nodeCategoryobjectBut the tradeoffs are worth it, especially because those performance hits can probably be optimized away in the future.
This overlay/underlay system will allow us to get simpler fine grain control over each field. We no longer need to do a search for a particular field, such as category's "name" field, we can just parse the whole category node using the normal attribute parsing tech and then access the value from a member variable. EG: We were previously looking for just the "name" attribute, but now we can look for any xml tag that we have associated with the 'name' field, such as 'bh-name'. All without adding custom logic to the parser.