gcl-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gcl-devel] Many warnings removed for Windows port using gcc 3.2 - U


From: Mike Thomas
Subject: Re: [Gcl-devel] Many warnings removed for Windows port using gcc 3.2 - Unix people beware
Date: Tue, 24 Sep 2002 16:10:10 +1000

Hi Dave.
 
Thanks for the feedback.  I am planning to use 3.2 at home and 2.95 at work to make sure that version dependencies don't creep in.  Probably not for much longer though as the new Mingw2 package is 3.2 anyway, and it is a convenient installlable exe for MinGW beginners to use.
 
I get the same results as you do below using gcc 2.95.  I worked around the include problem by doing:
 
export C_INCLUDE_PATH="g:/gclm/lib/gcl-2.5.0/h/"
 
before typing "make".
 
The include path problem expresses itself for me as follows:
 
-----------------------------------------------------------------
....
Error: (SYSTEM "gcc -c -Wall -fwritable-strings -DVOL=volatile -O2 -fsigned-char  -O2 -c -w binary-gcl/sloop.c -o binary-gcl/sloop.o") returned a non-zero value 1.
....
-----------------------------------------------------------------
 
The include path has not been set via a "-I" option - there is no way gcc can know where to look otherwise.  I think that this call is in the Maxima defsystem code, but I won't have time to look at how it is constructed until later this evening (if then).
 
XMaxima is also doing well under the other Mike's care and attention - I think I cc'ed a test result to the Maxima list this afternoon.
 
Cheers
 
Mike Thomas.
 
 
 
----- Original Message -----
Sent: Tuesday, September 24, 2002 3:24 PM
Subject: RE: [Gcl-devel] Many warnings removed for Windows port using gcc 3.2 - Unix people beware

OK.  Same result - CVS gcl builds OK on win2k using gcc-3.2.  I saw the same problem with
cmpinclude.h, which I "fixed" by copying cmpinclude.h into the ming/include directory.
I suspect that there are some (gcc or msys) path issues to fix here.
 
cvs maxima builds, but there are two new testsuite failures (below).  Stupidly I didn't check
to see if gcc-2.95 gave the same result (and I blew that compiler away).  I will reinstall and
try again later.
 
Error(s) found in rtest6a.mac: (4)
Error(s) found in rtest9.mac: (61)
/* ********************** Problem 4. *************** */
%Input is
ASSUME(A > 0)
 

The result is
[A > 0]
 
Using ratsimp
Using ratsimp
This differed from the expected result:
[a > 0]
/* ********************** Problem 61. *************** */
%Input is
 SIN(A X)
RESIDUE(--------, X, 0)
     4
    X
 

The result is
  1
- -
  6
 
Using ratsimp
Using ratsimp
This differed from the expected result:
   3
- A
----
 6
-----Original Message-----
From: Mike Thomas [mailto:address@hidden
Sent: Tuesday, 24 September 2002 1:01
To: address@hidden
Subject: [Gcl-devel] Many warnings removed for Windows port using gcc 3.2 - Unix people beware

Hi there.
 
I made many changes today mainly to remove hordes of compiler warnings under Windows.  There are still quite a few to go.  I can't guarantee that I have not accidentally trodden on other platforms so please check.
 
GCL builds with the new Mingw 2 package (uses gcc 3.2) provided that optimisation is set to "-O2" and "-fomit-frame-pointer" is omitted.  Maxima doesn't build because for some reason or other "cmpinclude.h" is not found (when compiling "sloop.lsp"), even though it was installed.
 
Anyway - 1 AM is too late for me so I'm quitting for the day.
 
Cheers
 
Mike Thomas.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]