RE: RE: utf-8 test diagram

["Maiorana, Jason" <jmaiorana idirect net>]

>To display glyphs on a screen, you need that the whole process
>is able to process said glyphs (and much more: xfs, X11, etc.).
Currently,
>dia 0.90-RC# is able to process UTF-8 data; however, one of its modules
>can't. That module is called gtk.

ok, thanks for the reply Cyrille.

I was suspicious that gtk was the problem, and that gtk2 will be the
answer.
hopefully that also means we can get some alpha and antialiased
fonts(xft),
in addition to unicode support in the gui.

the whole gnome2 gdk/gtk2 etc platform seems to have many features that
are fundamental afaic, so Im glad to see we're finally getting near
release.

Ill refrain from unicode/utf-8 questions until after the first gtk2
release,
but i still think its time to drop the whole # delimited string thing :)

(still support loading old diagrams though) I dont think this will
require
a format distinguisher in the file, or a new DTD (is there really one?)

maybe just a new end to data_string:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
--- dia_xml.c   Thu May 23 12:55:16 2002
+++ new.dia_xml.c       Thu May 30 11:41:42 2002
@@ -622,19 +622,23 @@

   if (data->xmlChildrenNode!=NULL) {
     p = xmlNodeListGetString(data->doc, data->xmlChildrenNode, TRUE);
-
-    if (*p!='#')
-      message_error("Error in file, string not starting with #\n");
-
-    len = strlen(p)-1; /* Ignore first '#' */
-
+
+    if(!p) return 0;
+
+    len = strlen(p);
+
     str = g_malloc(len+1);

-    strncpy(str, p+1, len);
-    str[len]=0; /* For safety */
+    if(str) {
+        if( len>=2 && p[0]=='#' && p[len-1]=='#' ) {
+           /*hash delimited string*/
+           strncpy(str, p+1, len-2);
+           str[len-2] = 0;
+        }
+        else
+           strncpy(str, p, len+1);
+    }

-    str[strlen(str)-1] = 0; /* Remove last '#' */
-
 #ifdef UNICODE_WORK_IN_PROGRESS
     return str;
 #else
@@ -643,7 +647,7 @@
     return str2;
 #endif
   }
-
+
   return NULL;
 }
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

maybe it doesnt matter, does anyone else think it would be nicer
not to have #'s? (esp. those who parse dia files and turn them
into other things...)