Reading Windows text file in Fortran?

Question

Is there a way to make the f90 compiler read Windows text files properly? I made a simple test with a text file created on a Windows system containing these three lines:
This is a Windows text file.
It has some extra control characters.
Unix does not have those.
The output I got was:
<-ext file.
<-e extra control characters.
<-se.
Kevin

Answer 1

That makes more sense. The problem isn't with the Fortran program, but with the Unix terminal (xterm, or whatever you're using to display the output). If you pipe the output through od like this you'll see all the characters, including the ^M:
bash-3.00$ fortranread | od -c
0000000       w   i   n   f   i   l   e       i   s       -   >   T   h
0000020   i   s       i   s       a       W   i   n   d   o   w   s   
0000040   t   e   x   t       f   i   l   e   .  \r                   
0000060                                                               
0000100                                           <   -  \n       w   i
0000120   n   f   i   l   e       i   s       -   >   I   t       h   a
0000140   s       s   o   m   e       e   x   t   r   a       c   o   n
0000160   t   r   o   l       c   h   a   r   a   c   t   e   r   s   .
0000200  \r                                                           
0000220                               <   -  \n       w   i   n   f   i
0000240   l   e       i   s       -   >   U   n   i   x       d   o   e
0000260   s       n   o   t       h   a   v   e       t   h   o   s   e
0000300   .  \r                                                       
0000320                                                               
0000340                   <   -  \n
0000347When the terminal sees the ^M character, it does a carriage return. That means that the next character gets written in the first column. Since the characters after the ^M are spaces, that makes the beginning of the lines appear blank.
If you want to see the output you expect, you need to pipe it through "tr -d '\015'", or you need to make the Fortran program itself delete the ^M characters. For example, you could write "myline(1:len_trim(myline)-1)" to always strip the last character of the line.
We should consider making formatted reads strip the ^M by default. I don't know whether any program would actually want them.