Re: Bidi issues
"Adam M. Costello" <[email protected]>
| Newsgroups | gmane.ietf.imaa |
|---|---|
| Message-ID | <[email protected]> |
Clearly we have an open issue that wasn't identified in the draft. Roy, thanks for spotting this. Martin Duerst <[email protected]> wrote: > there is no need to apply the bidi check to the whole left hand part > if that's not deemed appropriate. Okay, so we do indeed need to make a judgement call. In the current IMAA draft, Nameprep is applied to the entire local part before the local part is divided into segments. There is an important reason for that: the normalization step of Nameprep might cause the appearance of protected characters that weren't present before. Protected characters delimit segments. If we did segmentation first, then applied Nameprep to each segment, we might end up with segments containing protected characters, which defeats the whole point of segmentation (which is to make every segment contain no protected characters or to contain only protected characters, so that protected characters cannot get moved around by Punycode). If we don't want the bidi check to apply to the whole local part, but rather to individual segments, then I think we'll need to break Nameprep into two halves. The order of processing would be: Nameprep first half (mapping, normalization, prohibition) Segmentation Nameprep second half (bidi check) applied to each segment Existing implementations of Nameprep would need to change their interfaces to accomodate this. The alternative is to forbid the mixture of left-to-right and right-to-left characters in a local part (as the current draft does). Roy Badami <[email protected]> wrote: > if IMA practice were to follow the currently proposed IRI practice (ie > render by simply applying bidi in an LTR context), then the e-mail > address (logical order) [email protected] would render as: > > [email protected] That looks shocking to me, but it's consistent with the IRI draft section 4.4 example 5: logical: http://ab.cd.EF/GH/ij/kl.html visual: http://ab.cd.HG/FE/ij/kl.html The domain label and path component are swapped, just as the local part and domain label are swapped in your example. > Further, consider the address [email protected] (logical order). That's a good example, because the current draft prohibits it (because "owner" is left-to-right and "LIST" is right-to-left). If we want to allow this local part to exist at all (regardless of how it should be displayed), we'll need to change something. AMC