Let me start by saying that Gavin is absolutely correct.
However, if fixing it Gavin's way will take more than an hour or so, I would suggest that we land Andres's fix first, and then prioritize Gavin's follow-on fix as a Medium/Low. I think there are more important bugs to work on right now.
Reasoning: it's unlikely that the AMT will return XML documents in inconsistent encodings. Since this is a power driver that is very specific to the AMT, it is likely to work for the foreseeable future if we hard-code the encoding we know the AMT is using today. (if it's using UTF-8, for example, it is unlikely to change for the lifetime of the AMT product, IMHO.)
Let me start by saying that Gavin is absolutely correct.
However, if fixing it Gavin's way will take more than an hour or so, I would suggest that we land Andres's fix first, and then prioritize Gavin's follow-on fix as a Medium/Low. I think there are more important bugs to work on right now.
Reasoning: it's unlikely that the AMT will return XML documents in inconsistent encodings. Since this is a power driver that is very specific to the AMT, it is likely to work for the foreseeable future if we hard-code the encoding we know the AMT is using today. (if it's using UTF-8, for example, it is unlikely to change for the lifetime of the AMT product, IMHO.)