On 6/12/12 1:44 PM, Justin Karneges wrote: > On Tuesday, June 12, 2012 12:10:30 PM Peter Saint-Andre wrote: >> On 5/24/12 8:44 AM, Peter Saint-Andre wrote: >>> On 5/24/12 1:01 AM, Sergey Dobrov wrote: >>>> On 05/23/2012 04:45 AM, Peter Saint-Andre wrote: >>>>> Old thread alert! >>>>> >>>>> >>>>> Well, the spec says that the length restriction applies to the >>>>> base64-encoded data, so I think we meant that this does *not* apply to >>>>> the raw data. However, you seem to be saying that this might be >>>>> difficult to implement. Have other folks experienced this problem? >>>> >>>> Have other folks experienced this problem implementing things right ;) >>>> >>>> I know that some implementations (maybe all?) measure raw data and not >>>> base64. >>> >>> If that's what all code does now, then I think it's fine for us to >>> update the spec so that it reflects reality. >> >> Does anyone object to making this change in the spec? > > I think measuring the raw (pre-base64-encoded) data is the most sensible, and > likely what I had originally intended but poorly conveyed in the XEP. So this > is fine by me, especially if implementations are working this way.
Super. I propose the following change. OLD The base64-encoded data to be sent, prior to any wrapping in the <data/> element and IQ or message stanza, MUST NOT be larger than the 'block-size' determined in the bytestream negotiation. NEW The data to be sent, prior to base64-encoding and prior to any wrapping in XML, MUST NOT be larger than the 'block-size' determined in the bytestream negotiation. /psa
