Gelex problem

Here is a section of a .l file: 
 

 <IN_CADL_SECTION>{
 ^[ \t]+[^\n]*\n {
 io.put_string ("##### matched whitespace non-lf* lf #########%N")
 io.put_string ("%T##### " + text + " #########%N")
 in_buffer.append_string(text)
 in_lineno := in_lineno + 1
 }
 \n+ in_lineno := in_lineno + text_count
 

 ^[^ \t] { -- non-white space at start
 io.put_string ("##### matched non-whitespace at start #########%N")
 io.put_string ("%T##### " + text.item(1).code.out + " #########%N")
 unread_character(text.item(1))
 last_token := V_CADL_TEXT
 create str_.make (in_buffer.count)
 str_.append_string (in_buffer)
 in_buffer.wipe_out
 last_string_value := str_
 set_start_condition(INITIAL)
 }
 }

 

 

 The 3rd pattern (^[^ \t]) matches the first character on a line of input text that starts with 4 spaces and generates the following:
 

 
 ##### matched non-whitespace at start #########
         ##### 32 #########
 

 

 but the pattern is looking for non-whitespace at line start i.e. the [^ \t] bit. But 32 is a space.
 

 I had the impression this always worked in my system. But perhaps I have got something wrong in either the regex or the sub-state section. I double checked that the spaces in the regexs really are spaces and not TABs.
 

 

 However, if the line of input text has TABs not spaces, the parser works as expected. 
 

 Any ideas?
 

 - thomas