OCamlSDL

Everything compiled fine, except the
sdlevent_stub.c. The same syntax error as before. At
lines 329, 362 and 394, where you have

  SDL_Event *evt = stat_alloc(n * sizeof SDL_Event);

there should be parentheses around SDL_Event for the
sizeof:

  SDL_Event *evt = stat_alloc(n * sizeof(SDL_Event));

this solves the problem. 


---
[]s, Andrei de A. Formiga


--- Olivier Andrieu <oliv__a@...>
wrote:
> Hi,
> 
> I've committed code in the CVS that now uses alloca
> instead of
> malloc/free (GCC still use variable length arrays)
> in the C
> stubs. 
> 
> Andrei, could you have a look and tell me if it
> works with MSVC ?
> 
> -- 
>    Olivier
> 
> 
>
-------------------------------------------------------
> This SF. Net email is sponsored by: GoToMyPC
> GoToMyPC is the fast, easy and secure way to access
> your computer from
> any Web browser or wireless device. Click here to
> Try it Free!
>
https://www.gotomypc.com/tr/OSDN/AW/Q4_2003/t/g22lp?Target=mm/g22lp.tmpl
> _______________________________________________
> ocamlsdl-devel mailing list
> ocamlsdl-devel@...
>
https://lists.sourceforge.net/lists/listinfo/ocamlsdl-devel


__________________________________
Do you Yahoo!?
Protect your identity with Yahoo! Mail AddressGuard
http://antispam.yahoo.com/whatsnewfree

Francesco,

If you don't get the error at link time then I really think that godi=20
runs ranlib automatically when it installs files on macosx (maybe godi=20=

has some kind of verbose mode where you can check that ?). I applied=20
your godi-patches one by one to a fresh ocamlsdl-0.7.1, installed it=20
via a sudo make install and I get the error previously mentionned at=20
link time. This is not surprising if you refer the to bug described in=20=

man ranlib.

Le 25 janv. 05, =E0 01:37, Olivier Andrieu a =E9crit :

> OK here's a patch that does a ranlib after the install. Does this need
> to be done on the ocamlopt-generated .a files too ?

Apparently yes. After a make install, the error also appears at link=20
time for e.g. sdl.a when I try to build an example.

Daniel=

> Daniel B=FCnzli [Tue, 25 Jan 2005]:
 > > OK here's a patch that does a ranlib after the install. Does this
 > > need to be done on the ocamlopt-generated .a files too ?
 >=20
 > Apparently yes. After a make install, the error also appears at
 > link time for e.g. sdl.a when I try to build an example.

OK, so here's an updated patch that does the ranlib on all .a
files. If it works for you and Francesco, I'll commit it.

--=20
   Olivier

My sdl libs are installed in /opt/local/lib,

Trying to configure fails with either of the following invocations :

 > ./configure
 > env LDFLAGS=-L/opt/local/lib ./configure

[...]
checking for SDL_ttf.h... yes
checking for TTF_OpenFontRW in -lSDL_ttf... no
checking for TTF_OpenFontIndex in -lSDL_ttf... no
checking for TTF_OpenFont in -lSDL_ttf... no
configure: error: Cannot find SDL_tff library.


I attached the log for the second invocation at the end. Configure 
succeeds only
if you configure with

 > env LDFLAGS='-L/opt/local/lib -lobjc' ./configure

Note that with Francesco's patch everything was configuring well with a 
simple

 > ./configure

Thanks for your work,

Daniel

> Note that with Francesco's patch everything was configuring well with a 
> simple

It seems to me that Olivier's patch did not include the updated
configure.  Can you apply Olivier's patch on a clean 0.7.1, run
autoconf, and then configure ; make and tell us what you get?  Many
thanks (I do not have my laptop with me now).

-francesco

> Francesco Zappa Nardelli [Thu, 27 Jan 2005]:
 > 
 > > Note that with Francesco's patch everything was configuring well
 > > with a simple
 > 
 > It seems to me that Olivier's patch did not include the updated
 > configure.  Can you apply Olivier's patch on a clean 0.7.1, run
 > autoconf, and then configure ; make and tell us what you get?  Many
 > thanks (I do not have my laptop with me now).

Ah yes : my patch is against the CVS and configure is not in the CVS
since it's a generated file.

-- 
   Olivier

Le 27 janv. 05, =E0 15:33, Francesco Zappa Nardelli a =E9crit :

>
>> Note that with Francesco's patch everything was configuring well with=20=

>> a
>> simple
>
> It seems to me that Olivier's patch did not include the updated
> configure.  Can you apply Olivier's patch on a clean 0.7.1, run
> autoconf, and then configure ; make and tell us what you get?  Many
> thanks (I do not have my laptop with me now).
Right. Configure works fine. Build works fine.

However for install, the ranlib executions do not take place. The=20
reason is that $(PLATFORM) contains the value "@PLATFORM@", instead of=20=

"Apple". In makefile.config.gcc the value of $(PLATFORM) is=20
"@PLATFORM@", in makefile.platform the value of $(PLATFORM) is "Apple"=20=

but in the toplevel Makefile we include the latter before the former.=20
So "@PLATFORM@" is taken as the value for $(PLATFORM).

Daniel=

> Daniel B=FCnzli [Thu, 27 Jan 2005]:
 >=20
 > Le 27 janv. 05, =E0 15:33, Francesco Zappa Nardelli a =E9crit :
 >=20
 > >
 > >> Note that with Francesco's patch everything was configuring well
 > >> with a simple
 > >
 > > It seems to me that Olivier's patch did not include the updated
 > > configure.  Can you apply Olivier's patch on a clean 0.7.1, run
 > > autoconf, and then configure ; make and tell us what you get?
 > > Many thanks (I do not have my laptop with me now).
 > Right. Configure works fine. Build works fine.
 >=20
 > However for install, the ranlib executions do not take place. The
 > reason is that $(PLATFORM) contains the value "@PLATFORM@", instead
 > of "Apple". In makefile.config.gcc the value of $(PLATFORM) is
 > "@PLATFORM@", in makefile.platform the value of $(PLATFORM) is
 > "Apple" but in the toplevel Makefile we include the latter before
 > the former.  So "@PLATFORM@" is taken as the value for $(PLATFORM).

Ah right, simply remove the PLATFORM line in makefile.gcc.in and it
should work.

--=20
   Olivier

Le 27 janv. 05, =E0 18:47, Olivier Andrieu a =E9crit :

> Ah right, simply remove the PLATFORM line in makefile.gcc.in and it
> should work.
Ok works fine here with that fix.

Daniel


P.S. As stated in my original message, you should add this somewhere to=20=

warn osx users and avoid messages to ocamlsdl-users :

(warning, same information is replicated in INSTALL and README)

Using
=3D=3D=3D=3D

1. Native code. Same as in the INSTALL/README.

2. Byte code.
You cannot use dynamic linking (why ? see explanation of the patch),=20
you need to use -custom while linking.

 > ocamlc -I +sdl -c my_prog.ml
 > ocamlc -I +sdl -custom -o my_prog bigarray.cma sdl.cma my_prog.cmo

3. Top-level
Here again you cannot use dynamic linking you need to build a custom=20
top level.

 > ocamlmktop -I +sdl -custom -o ocamlsdl bigarray.cma sdl.cma ...
 > ./ocamlsdl -I +sdl

Note that in (2) and (3), if dynamic linking is used, the result is=20
that a bunch of errors is printed on stdout and the process is aborted=20=

(but it doesn't segfault).=

> Daniel B=FCnzli [Thu, 27 Jan 2005]:
 >=20
 > Le 27 janv. 05, =E0 18:47, Olivier Andrieu a =E9crit :
 >=20
 > > Ah right, simply remove the PLATFORM line in makefile.gcc.in and
 > > it should work.
 > Ok works fine here with that fix.
 >=20
 > Daniel
 >=20
 >=20
 > P.S. As stated in my original message, you should add this somewhere t=
o=20
 > warn osx users and avoid messages to ocamlsdl-users :
 >=20
 > (warning, same information is replicated in INSTALL and README)

OK ! I committed the patch, added a README.macosx file with this
information, removed the duplicate info in README and INSTALL, and made
a ocamlsdl-0.7.2 release with it.

Non-MacOSX users need not upgrade to this release as there is no
change other than the build system improvements for MacOSX.

--=20
   Olivier