"Remko Lodder" <remko_(_at_)_elvandar_(_dot_)_org> wrote in <64186_(_dot_)_145_(_dot_)_221_(_dot_)_92_(_dot_)_42_(_dot_)_1131027390_(_dot_)_squirrel_(_at_)_webmail_(_dot_)_evilcoder_(_dot_)_org>: re> Not by sgmlnorm indeed, but a newer version of tidy is capable of doing re> so. Hiroki Sato was looking into this and it's one of my open PR's.... re> (hrs CC'ed) re> re> What was said at the time that we only needed to merge the --preserve re> option and test away after that. So we can solve this, not by sgmlnorm re> but via tidy. The only difficult item is that there is no normal re> version of tidy anymore. The current tidy is not maintained anymore and re> there is a devel version available which updates every x period of re> time. re> re> @hrs: idea's? Sorry for the delay. This case conversion of the id attribute is required according to the SGML specification, so the sgmlnorm's behavior is correct. It means <foo id=bar> and <foo id=BAR> are identical and sgmlnorm normalizes them into <foo id=BAR>. You can see attributes other than "id" preserves the case. However, the XML specification distinguishes the case. We have used HTML 4.01 DTD (SGML) for our www tree and *forcibly* converted the docs to XHTML 1.0 (XML) by using tidy(1). This problem is caused by this pushiness. Well, I think simply using capital letters in the id attribute consistently solves the problem...or, am I missing something? -- | Hiroki SATO
Attachment:
pgpq3dwd9qMoI.pgp
Description: PGP signature