The "synchronous" part of the URLFetcher works very differently from the "asynchronous" part and is very buggy. The problems include:
  - In case of a synchronous fetcher, the initConn() method is not called and thus the connection is not properly setup wrt to timeouts, accepted formats (gzip), and the use of caches
  - Errors are reported inconsistently
  - In case of error, the synchronous fetcher does not close the error stream
It would perhaps be ideal to unify the synchronous and asynchronous fetchers by having the synchronous loading go through mostly the same code path as the asynchronous one.