g77 ? why conflicting REC= and END= in read ?

[Dave Love]

>>>>> "PT" == Peter Teuben <teuben@astro.umd.edu> writes:

 PT> Why would g77 flag the following code as wrong:

[...]

 PT>                read(11,rec=i,end=800, err=999) j
 PT>                        1     2
 PT> Conflicting I/O control specifications at (1) and (2)


 PT> (and btw, the presence of the err= is not important, but for
 PT> historic (?) reasons this code needs both err= and end=.

Huh?  What does the code expect this to do?

 PT> I looked over the ANSI manual, 

Good!

 PT> can't find any restrictions on the combined use of END= and REC=,

Section 12.8.1.

 PT> and the code compiles fine on a variety of other compilers.

and doesn't on others.