Fair enough. You've won this week's prize of affection from the editors. :-) JJ. Jacek Kopecky wrote: > Jean-Jacques, > as Jonathan has already mentioned, I believe, one can have a safe > operation that uses POST. > > Safety is used for defaulting to GET so that you can have interfaceless > bindings that nevertheless use GET for some operations. Such defaulting > may even be incorrect in light of how much request data can be > represented usefully and interoperably in a URI. There are practical > limits to that. > > I don't think it's useful to tie {http method} and {safe} more than they > are tied now. > Best regards, > Jacek > > > > > > On Mon, 2007-02-12 at 14:51 +0100, Jean-Jacques Moreau wrote: > >> I've implemented the current resolution for CR123. >> >> However, I think we need to do a little more. Indeed, currently: >> - if both {http method} (or, in its absence, {http method default}) and >> {safe} are set, {http method} takes precendence, i.e. the value of the >> {safe} property is ignored. >> >> This could be an issue, for example if {http method}="POST" and >> {safe}="true". >> >> Two alternatives: >> a. Indicate {http method} and {safe} cannot be present simultaneously. >> >> b. {safe} takes precendence over {http method} (as {http method} already >> takes precendence over {http method default}). >> >> There's some appealing symmetry in b., but a. may be straigther. >> >> JJ. >> >> > > >Received on Monday, 12 February 2007 17:33:46 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 12 February 2007 17:33:47 GMT