> Ok.  So what we're specifically talking about is this, right:
> Index: includes/http_header.inc
> ===================================================================
> --- includes/http_header.inc  (revision 1172)
> +++ includes/http_header.inc  (working copy)
> @@ -1,7 +1,7 @@
> "http://www.w3.org/TR/html4/strict.dtd";>
>  <HTML>
>  <HEAD>
> -  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
> +  <META http-equiv="Content-Type" content="text/html; charset=UTF-8">
>  <?php
>  if (isset($meta_desc)) {
>      print("<META name=\"description\" content=\"$meta_desc\">\n");
> But even with this change, it's still safest to use the &# encoding of the 
> Hebrew characters in mirrors.txt, right?  Or does this change allow me to put 
> back the Hebrew characters in mirrors.txt?  Which way is preferred, or are 
> they more-or-less equivalent?
> As a side note: I'm wondering how you come up with the &#... encoding for a 
> random set of non-english characters.  Are there UTF-8 -> &#encoding 
> convertors or something?

  Yes - That should do it. It's still safer to use the &# encoding as then
  even if they don't have UTF-8, but do have the required font it would
  still work. As for the entities there is a giant list you can look at, 
  or often tools like http://www.wandel.person.dk/hebrew.html out there 
  written by some bored person. 

  Personally I think strait UTF-8 is "cleaner" but the entities are most
  likely the "best" way for the lowest common demoninator, if not the right
  way to do it. 

  Reference Documents: 
        "When to use Escapes" 

        UTF-8 Table

  Most likely more then you needed, but I've spent so much time fighting
  character set issues they've become somewhat of a passion... 

> FWIW, we use a postgres DB for our regression testing database (see 
> www.open-mpi.org/mtt/), but we don't allow that stuff to be mirrored.  
> There's waaaay too much data accumulated each night, and it would also 
> significantly raise the bar on the requirements for mirroring the Open MPI 
> web site.  Not desirable.  :-)

  Hehe - Yeah no thanks! :)

-Karl Vollmer 

Reply via email to