Joe Gregorio wrote: > On 1/28/06, Sam Ruby <[EMAIL PROTECTED]> wrote: > >>James M Snell wrote: >> >>>This pace requires that a server either accept >>>everything that is given to it or declare exactly what it will accept up >>>front. >> >>If this pace is scheduled, I will -1 it. >> >>The APP needs to take a "Small Pieces, Loosely Joined" approach. >> >>A server may reject entries that contain pub:slugs that are 32,768 bytes >>long, or contain text that is critical to the Chinese government, or >>contain entirely well-formed and valid entries that are encoded in EBCDIC. >> >>And yet, very few of us would ever be affected by any such restrictions. >> >>Similarly, very few of us would be affected by a site that prohibit >>markup in titles. > > That depends, what do you mean by prohibit? > That there is silent data loss, i.e. the server > strips markup from titles? > > Or do you mean that the server sends a 4xx > response with an error report that the server > does not accept entries with markup? In this > case you are shifting the problem from > the client developer's shoulders onto the > client users' shoulers.
Perhaps I should have put the answer earlier in the post. ;-) First repeating: > In my opinion, it is probably worthwhile for the server to indicate that > markup is not allowed in titles, and this may in fact be a common enough > requirement that it should be a standard extension Then expanding: Clients that don't look for this will be less easy to use in some edge cases (I've *never* put markup in titles, have you?, and I'm not talking about plain text like Tim's classic "</tag>" post which was interpreted as markup insteed of being plain text, but fonts, links, or bold, or italics). While this restriction has gotten a lot of dicussion as it is amoungst the first we have encountered, I expect that there will be a lot of restrictions - enough so that we can't possibly enumerate them all (examples: a server that won't less you post adult content, or contents that aren't well formed (I'm thinking Jacques Distler and Tim Bray here), limit the number of links (APP can be used for comments, right?, see Rogers Cadenhead's policy), etc. Thinking about it, the extension mechanism may need to have a standard means for indicating human readable version of the restriction, as there may be some cases where expecting the client to enforce it isn't practical. - Sam Ruby
