- From: Dmitry Turin <html60@narod.ru>
- Date: Fri, 18 May 2007 15:58:11 +0300
- To: public-html@w3.org
Good day, Dao. Excuse me for a long break. DG> <style> DG> a::before { DG> content: attr(a1) "\A" attr(a2) "\A" attr(a3) "\A" attr(a4); DG> } DG> </style> <style> record1::before { content: attr(name) attr(surname) } record2::before { content: attr(name) attr(surname) } </style> <body> <record1 name= surname= > <record2 name= surname= > </body> This way don't allow to specify styles for visualization of these attributes (like the following) <style> record1 §name, §surname {color: red } record2 §name, §surname {color: gray } </style> <body> <record1 name= surname= > <record2 name= surname= > </body> </style> So we see, that michanism of property "content" is _essentially_ narrower, than michanism of virtual tags. Virtual tag allow to use even pseudo-classes. I don't want to reduce importance of property "content": it is necessary to change already existing content of html-element, or to specify content of pseudo-element. But when you try to reduce all cases to property "content", property acts as limiter, and thus data are subdivided into two groups depending on their forms: if data is between an open and close tags, it is data _of full value_ (it may has style); if data is in an attribute, it is _inferior_. It's not reasonable to abandon virtual tags, only because more weak decision was already thought up. P.S. Other example <style> tab { display: table } a { display: table-row } a §* { display: table-cell } </style> <body> <tab> <a a1="v11" a2="v12" a3="v13"> <a a1="v21" a2="v22" a3="v23"> <a a1="v31" a2="v32" a3="v33"> </tab> </body> Dmitry Turin http://html6.by.ru http://sql4.by.ru http://computer2.by.ru
Received on Friday, 18 May 2007 12:58:30 UTC