"Dmitrij D. Czarkoff" writes: > Sorry for replying to single message with two. > > Anthony J. Bentley said: > > HTML5 has been some steps forward and some steps back. But one of the > > unambiguously good things they did was drop any pretense of SGML > > compatibility, and introduce well‐defined error handling rules (instead > > of the XML practice of dropping things on the floor as soon as it sees > > a missing angle bracket). > > IMO this is the worst thing about HTML currently. There may be only > three possible rules for sane markup language: > > * drop offendig subtree, > * abort rendering on first error or > * replace malformed document with warning.
How many Unix tools behave this way? Does grep abort upon encountering invalid UTF-8 sequences in a file? No. Does troff abort rendering on invalid macro usage? Practically never. -- Anthony J. Bentley
