format interpreter - bug ?

Luca Bottura
Wed Apr 14 06:53:00 GMT 1999

I have isolated the following anomalous response to egcs to an error in
the specification of an implicit format. The program:

      real a(20)

      do i=1,20
         a(i) = 1
      write(6,'(10((1x,1p,e11.3))') (a(i),i=1,20)


compiles and links in spite of the wrong number of parentheses in the
format (three openj parentheses, two closed parentheses). At runtime the
executable crashes (as expected)  with the error:

startio: error in format
apparent state: (10((1x,1p,e11.3))
lately writing sequential formatted external IO
Aborted (core dumped)

Why is the format error not caught before (at compilation) ?

I am running egcs - 1.0.2 prerelease on MkLinux DR3.0 (calling egcs
--version results in the output of egsc - 2.90.25 980302 egcs - 1.0.2

tqvm, Luca

More information about the Gcc-bugs mailing list