| Re: [AD] Allegro 4.2.1 stress test |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
Andrei Ellman wrote:
However, one thing I've noticed about fix.sh: fix.bat includes separate
options for MSVC6, MSVC7, and MSVC8, whereas fix.sh is just MSVC. Do the
extra options in fix.bat that are MSVC7 or MSVC8 specific do anything
that would be useful in a UNIX-like environment (like Cygwin)? The MSVC6
Allegro builds fine under Cygwin, but I would recommend that someone
tests building Allegro for MSVC7 and MSVC8 using Cygwin.
Yes, fix.sh misses support for MSVC7&8, but that's not a problem since
you can run fix.bat from cygwin. I built allegro for MSVC8 using cygwin
just fine.
Also, has
anybody tested with MinGW (both with and without MSYS) (both the MinGW
build and the MSVC build)?
Yes. I successfully built a MinGW version (without MSVC). There was a
problem with ALLEGRO_USE_C defined though. Compilation failed because
_stub_bank_switch, _stub_unbank_switch and _stub_bank_switch_end were
double defined. And indeed, makefile.mgw says:
...
ifdef ALLEGRO_USE_C
# ------ build a C-only version ------
VPATH += src/c
MY_OBJECTS = $(C_OBJECTS) cmiscs
CFLAGS += -DALLEGRO_NO_ASM
else
...
Bank switching routines are defined in both src/c/cmisc.c and
src/c/cmiscs.s. I tried to remove the "cmiscs" from "MY_OBJECTS =
$(C_OBJECTS) cmiscs" and it compiled fine, but example programs run with
a black screen. I have no idea what could be wrong and why is cmiscs.s
added to the C-only build. The comment in cmiscs.s is quite confusing:
"Empty bank switch routines for DOS version. Should be used together
with C versions of GFX_VTABLE routines and ASM calling convention."
I find it a minor issue because ALLEGRO_USE_C option is not even
mentioned in mingw build instructions.
--
Milan Mimica
http://sparklet.sf.net