how about we just use a foreach on footers and then combine all of these in a single pass? Seems simpler.

i.e. something like

var mergedSchema = StructType(Nil)
footers.foreach { file =>
  val schema = ParquetRelation.readSchemaFromFooter(footer, converter)
  try {
    mergedSchema = mergedSchema.merge(schema)
  } catch {
    ...
  }
}
mergedSchema

Can we get the schema from first footer and then go through this loop for remaining footers? Because you merge the first schema with an empty schema, I think the all fields in merged schema will be optional in their metadata. So the pushing down of filters will not normally work.

Yeah you'll right, filter push-down can be affected due to #9940 (which I just merged today). Thanks for pointing this out!

can we catch something tighter here? This is too broad.

also i'd prefer to make this more concise, e.g.

                  throw new SparkException(
                    "Failed merging schema of file ${footer.getFile}: ${schema.treeString}", cause)

I don't know why these functions are super long.