Cart
[Home] [By Thread] [By Date] [Recent Entries]
Michael Kay writes: > ... > I suspect those who argued that terseness was not important for XML > were actually arguing that human readability is more important than > message size. For what it's worth, that matches my recollection. > ... > I also suspect that the reason the matter came up for debate was the > more specific question of whether element names should be repeated in > the end tag. To be honest, I'm still unsure in my own mind about that > decision -- there are arguments both ways. ... My memory, for what it is worth, is that the main reason the matter was explicitly addressed was the relatively large portion of ISO 8879 devoted to mechanisms for minimizing the number of keystrokes needed to encode a document. The SGML users in the working group were convinced by a large majority that mechanisms like SHORTTAG, OMITTAG, and DATATAG had unfavorable cost : benefit ratios: the level of complexity they introduced was not paid back by their usefulness. They were also a bit error-prone: even experienced SGML users reported discovering that their use of tag omission had led in some cases to document structures which were not quite what they had intended. So I read that item in the list of design principles as a signal: don't expect tag omission or any of the other paraphernalia of SGML that made it so hard for people to produce conforming SGML parsers. I have not checked the records of the WG, but I suspect that the decision on repeating the identifier in the end-tag came later and was handled by an appeal to the principle. Omission of the name works fine in short examples, but once the end-tag is more than a screen or two away from the start-tag, it's unhelpful. Making the name optional might have worked, but every optional feature adds complexity, and we had some very persuasive people arguing for cutting away every possible source of complexity. -- C. M. Sperberg-McQueen Black Mesa Technologies LLC http://blackmesatech.com
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] |
