comment in more detail about DC_DESIRED_TEXT_LEN and use old limit that had worked for some while okayish

src/constants.rs

@@ -168,12 +168,30 @@ pub const DC_MSG_ID_LAST_SPECIAL: u32 = 9;
 /// string that indicates sth. is left out or truncated
 pub const DC_ELLIPSE: &str = "[...]";
 
-/// to keep bubbles and chat flow usable, we limit the text length to DC_DESIRED_TEXT_LEN.
+/// to keep bubbles and chat flow usable,
+/// and to avoid problems with controls using very long texts,
+/// we limit the text length to DC_DESIRED_TEXT_LEN.
 /// if the text is longer, the full text can be retrieved usind has_html()/get_html().
-pub const DC_DESIRED_TEXT_LEN: usize = 20000;
+///
+/// we are using a bit less than DC_MAX_GET_TEXT_LEN to avoid cutting twice
+/// (a bit less as truncation may not be exact and ellipses may be added).
+///
+/// note, that DC_DESIRED_TEXT_LEN and DC_MAX_GET_TEXT_LEN
+/// define max. number of bytes, _not_ unicode graphemes.
+/// in general, that seems to be okay for such an upper limit,
+/// esp. as calculating the number of graphemes is not simple
+/// (one graphemes may be a sequence of code points which is a sequence of bytes).
+/// also even if we have the exact number of graphemes,
+/// that would not always help on getting an idea about the screen space used
+/// (to keep bubbles and chat flow usable).
+///
+/// therefore, the number of bytes is only a very rough estimation,
+/// however, the ~30K seems to work okayish for a while,
+/// if it turns out, it is too few for some alphabet, we can still increase.
+pub const DC_DESIRED_TEXT_LEN: usize = 29_000;
 
-/// approx. max. length returned by dc_msg_get_text()
-pub const DC_MAX_GET_TEXT_LEN: usize = 30000;
+/// approx. max. length (number of bytes) returned by dc_msg_get_text()
+pub const DC_MAX_GET_TEXT_LEN: usize = 30_000;
 
 /// approx. max. length returned by dc_get_msg_info()
 pub const DC_MAX_GET_INFO_LEN: usize = 100_000;