[Home] [By Thread] [By Date] [Recent Entries]
On Jan 2, 2007, at 17:11, Pete Cordell wrote: > In terms of end-of-line encoding, the approach seems to be to > output what is convenient (CR, LF, or CRLF) and have the receiving > application sort out the situation. More to the point, the LF character in element content can be serialized as CR, LF or CRLF. Of course, LF is the most natural serialization. In order to avoid dataloss, LF, CR and tab need to be escaped in attribute values. Otherwise they are normalized to space by the parser. This matters for example when round-tripping multiline values in XHTML <input type='hidden'/>. > Conceptually, the receiving XML processor should normalize the end- > of-line markers to 0x0A and then the application converts that to > which ever of CR, LF, or CRLF is appropriate. For this reason, in order to avoid dataloss, CR needs to be escaped as an NCR to make it survive serialization and parsing round-tripping. -- Henri Sivonen hsivonen@i... http://hsivonen.iki.fi/
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] |

Cart



