- From: Jacob Palme <jpalme@dsv.su.se>
- Date: Mon, 5 Apr 1999 12:44:54 +0200
- To: SeniorOnline technical mailing list <sol-tech@ambra.omega.it>
- Cc: IETF Applications Area Discussion List <discuss@apps.ietf.org>
At 10.33 +0200 99-04-05, Martin J. Duerst wrote: > This is more or less what HTTP has done. It prooved to be detrimental, > because ISO Latin 1 was used as a default (without explicit "charset" > paramenter) and in various parts of the world, there was a rather low > incentive to actually implement handling the "charset" parameter, or > adding one where it was needed. As a result, most HTTP trafic, whether > in ISO Latin 1 or not, is not correctly tagged. I think his arguments are rather valid. If we start using Unicode/ISO 10646 from the beginning, how much more difficulty will this cause in our development work (we have limited resources, being a publicly financed research project, not a commercial software development venture)? I can see two ways of doing this. Method 1: Use ISO Latin 1 internally, and just translate to UTF-8 before we send things using our new protocol, and translate back to ISO Latin 1 again the first thing when we get things back. This method would not cause much difficulty, just add two translation routines and some reduced CPU efficiency. However, this method is a kind of "fake" UTF-8, if someone tryes to send real UTF-8, containing other characters than those in ISO Latin 1 (such as for example Hungarian) then our software will not perform correctly. Method 2: Really use UTF-8 all the way interally, and send UTF-8 to the web browsers which people use as clients to Senior Online. This method will only work if all users have web browsers which support UTF-8. Another disadvantage with method 2 is that we will have to translate to ISO Latin 1 or some other simpler character set, when we send things via e-mail, since most e-mailers in the world do not support UTF-8 yet. --- --- If we do not use method 1 or 2, and if we start using only "ISO Latin 1", we should at least employ a charset parameter which is mandatory, and which does not default to ISO Latin 1. ------------------------------------------------------------------------ Jacob Palme <jpalme@dsv.su.se> (Stockholm University and KTH) for more info see URL: http://www.dsv.su.se/~jpalme
Received on Monday, 5 April 1999 06:51:07 UTC