Bring normalization algorithm closer to the spec
No logical difference so far.
This commit is contained in:
parent
e02d925786
commit
bec2ac4fde
|
@ -250,7 +250,8 @@ _hb_ot_shape_normalize (hb_font_t *font, hb_buffer_t *buffer,
|
|||
}
|
||||
|
||||
hb_codepoint_t composed, glyph;
|
||||
if ((buffer->out_info[buffer->out_len - 1].combining_class() < buffer->info[buffer->idx].combining_class()) &&
|
||||
if ((starter == buffer->out_len - 1 ||
|
||||
buffer->out_info[buffer->out_len - 1].combining_class() < buffer->info[buffer->idx].combining_class()) &&
|
||||
hb_unicode_compose (buffer->unicode,
|
||||
buffer->out_info[starter].codepoint,
|
||||
buffer->info[buffer->idx].codepoint,
|
||||
|
|
Loading…
Reference in New Issue