[ePiX-devel] Sample file failure with 1.1.12

[Andrew D. Hwang]

The sample file R_demo.xp won't compile for me in Version 1.1.12 under 
gcc-3.2 or gcc-3.3 on GNU/Linux, but will compile on Mac OS X with 
gcc-4.0. Here are the error messages from gcc-3.3.2:

epix-local-17447/R_demo-tmp.cc: In function `int main()':
epix-local-17447/R_demo-tmp.cc:55: error: syntax error before `,' token
epix-local-17447/R_demo-tmp.cc:55: error: no matching function for call to 
`ePiX::axis::axis(double&, double)'
/home/hwang/epix/pkg/epix-1.1.13/epix.h:1733: error: candidates are: 
ePiX::axis::axis(const ePiX::axis&)
/home/hwang/epix/pkg/epix-1.1.13/epix.h:1736: error: ePiX::axis::axis(const ePiX::P&, 
const ePiX::P&, unsigned int, const ePiX::P&, ePiX::epix_label_posn)
epix-local-17447/R_demo-tmp.cc:55: error: syntax error before numeric constant
epix-local-17447/R_demo-tmp.cc:56: error: request for member `subdivide' in 
`Ay', which is of non-aggregate type `ePiX::axis ()(...)'
epix-local-17447/R_demo-tmp.cc:58: error: no match for call to `(ePiX::axis) 
(double&, double)'
epix-local-17447/R_demo-tmp.cc:58: error: no match for call to `(ePiX::axis) 
(int, int)'

Line 55

  axis Ay(P(xmin(), 0), P(xmin(), 0.4), 4, P(-4,0), l);

is definitely the problem; removing everything else in the file preserves 
the error message. The following *fix* the problem:

1. Replacing the *first* call to xmin() with a double literal,
2. Replacing P(xmin(), 0) with a separately-constructed P, as in

  P tmp(xmin(), 0);
  axis(tmp, ...);

The call

  axis(P(0, ymin()), ...);

elsewhere in the file compiles fine, but fails with the same error if the 
components are switched. (It looks suspiciously like xmin() is returning 
an uninitialized reference, but the value is correct at the first call.)

The file works with Version 1.1.11. None of the recent changes seem 
obviously relevant to this snippet, but clearly at least one of them is. 
If necessary, I'll comb through change by change to see where the breakage 
occurred, but does anyone see why this call might suddenly start to fail? 
Specifically, have I made a coding gaff, or does this look like a compiler 
bug?

--
Andy

Andrew D. Hwang                 address@hidden
Department of Math and CS       http://mathcs.holycross.edu/~ahwang
College of the Holy Cross       (508) 793-2458 (Office: 320 Swords)
Worcester, MA, 01610-2395       (508) 793-3530 (fax)