ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Hi,

I just a preliminary look at trying to get DROD (Caravel) to compile on MacOS X. Needless to say the result was .. erm interesting.

I've just updated the source from CVS as per the recent changes on SGI, and there's no sign of a Makefile

I'm willing to help try and get this running on Mac (there is only so often my boss can catch me playing at lunchtimes ), which means compiling with GCC (which currently DOES NOT seem to like this code .. AT ALL ).

So my main question is, is there a makefile anywhere that could save me lots of time and effort working out what each directory builds into ??

mrimer
Level: Legendary Smitemaster

Rank Points: 5199
Registered: 02-04-2003
IP: Logged

ross wrote:
So my main question is, is there a makefile anywhere that could save me lots of time and effort working out what each directory builds into??

Uh, no. But it would be great to have someone help port this to the Mac! In short, each directory builds into a separate library (e.g., DRODLib) or executable. And the executables all depend on DRODLib.lib. And the source files all in all depend on the third party libs described in another thread in Development.

Hmm...maybe Adam or myself here would like to get DROD at least compiled on Linux, which shouldn't be too bad after Shick's Irix port. And from there it should be a quick leap over to Mac OS.

I think you'll want to use version 3 or higher of gcc, as it's more compliant to the C++ standard. Also, I'm sure we'll want to add some conditionally compiled statements for any gcc/OS specific stuff. Let's keep in touch with how our progress on this goes!

____________________________
Gandalf? Yes... That's what they used to call me.
Gandalf the Grey. That was my name.
I am Gandalf the White.
And I come back to you now at the turn of the tide.

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

mrimer wrote:
Uh, no. But it would be great to have someone help port this to the Mac! In short, each directory builds into a separate library (e.g., DRODLib) or executable. And the executables all depend on DRODLib.lib.

Good, that's where I started As soon as I get DRODLib built I'll let you know because I guess that will probably be the worst of it (hopefully).

mrimer wrote:
Hmm...maybe Adam or myself here would like to get DROD at least compiled on Linux, which shouldn't be too bad after Shick's Irix port. And from there it should be a quick leap over to Mac OS.

I didn't think it would be too bad getting the IRIX stuff onto OS X, but then I've only been trying for a half hour or so, so I wouldn't read anything into my comment

mrimer wrote:
I think you'll want to use version 3 or higher of gcc, as it's more compliant to the C++ standard. Also, I'm sure we'll want to add some conditionally compiled statements for any gcc/OS specific stuff.

I'm using 3.1
How is the IRIX code currently built ?

Schik
Level: Legendary Smitemaster

Rank Points: 5453
Registered: 02-04-2003
IP: Logged

ross wrote:
How is the IRIX code currently built ?

Ummm... with the Makefiles that I apparently never checked in.

And I just tried to connect to my SGI, seems its down right now (it's been having memory problems) and I won't be back at work to reboot it until Monday.

Sorry about that.


____________________________
The greatness of a nation and its moral progress can be judged by the way it treats its animals.
--Mahatma Gandhi

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

I can wait until Monday to play DROD on my mac ... no .. no I can't ... I wanna play now

Maybe I'll just have to pop into work tomorrow and get a fix then instead

Schik
Level: Legendary Smitemaster

Rank Points: 5453
Registered: 02-04-2003
IP: Logged

Okay man, you're SO excited to get on this, so I dropped by work to reboot my machine and get the files. They're checked in now. You better have this working by tomorrow!

Okay, okay, so I remembered I had a version on another machine too, and checked it in from home.

I'll warn ya - I haven't tried compiling on the SGI in a while, and there have been a LOT of changes. I've been meaning to do an update on my SGI and fix anything that's come up, but haven't had time to.

One thing you'll need to do is decide whether or not you can support Unicode. On the SGI I couldn't find functions to support it. Hopefully you'll be able to on the Mac. The HAS_UNICODE constant that you'll need to define (or not) is defined near the top of Files.h.

Good luck with it all, and feel free to ask if you need any help....

____________________________
The greatness of a nation and its moral progress can be judged by the way it treats its animals.
--Mahatma Gandhi

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Schik wrote:
Okay, okay, so I remembered I had a version on another machine too, and checked it in from home.

It's the thought that counts
i'm gonna play around for an hour or so until I have to go out (Archery competition - no chance of winning though) and then I'll finish what I can this evening. Fingers crossed

I'll warn ya - I haven't tried compiling on the SGI in a while, and there have been a LOT of changes. I've been meaning to do an update on my SGI and fix anything that's come up, but haven't had time to.

These Windows programmers eh
I'm expecting more than a few problems so no need to worry there. We'll see how it goes.

The biggest problem is that SDL is currently shipped as a Framework (don't ask) on MacOS X and there is a need for the main.cpp to be a .m file (Objective-C). I'm recompiling SDL from source as I type.

One thing you'll need to do is decide whether or not you can support Unicode. On the SGI I couldn't find functions to support it. Hopefully you'll be able to on the Mac. The HAS_UNICODE constant that you'll need to define (or not) is defined near the top of Files.h.

MacOS X *does* support Unicode, but I guess that #define HAS_UNICODE is going to throw me in with all the Win32 code and I don't want that I'll try without first and then see if I can add it later on.

Thanks for the quick reply, I'll keep you up to date on how it's going.

Update 1 ------------------------------
Got as far as EventHandlerWidget

There are LOTS and LOTS and LOTS of warnings. Mostly about friend declarations requiring a class key "friend X" should be "friend class X", although there are fair few about dodgy typecasting (unsigned int x = -1) etc.

I don't intend to fix this immediately unless it stops compilation, but does anyone else see this on their compiler (non-gcc) ?

Hopefully I can get it to compile (if not link) tonight.


[Edited by ross on 05-18-2003 at 10:09 AM]

[Edited by ross on 05-18-2003 at 10:09 AM]

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

I take it that this didn't compile under g++ on IRIX ?

I've got DROD itself compiling (but with hundreds of warnings) and I'm checking out DRODLib but the Makefile called from DROD/makefile.sgi isn't there. So for now I am using a copy of the DROD/Makefile.sgi.

If you want this to compile on Linux there are a lot of things that gcc doesn't like that will need changing.

Also Gamescreen.cpp line 1219 seems to a possible bug. Basically looks like
x = 1 ? y : z;
I assume it should be
x == 1 ? y : z;

I'm still chewing my way through ..

ErikH2000
Level: Legendary Smitemaster

Rank Points: 2796
Registered: 02-04-2003
IP: Logged

ross wrote:
Also Gamescreen.cpp line 1219 seems to a possible bug. Basically looks like
x = 1 ? y : z;
I assume it should be
x == 1 ? y : z;

const DWORD dwNextLevelID = pNextLevelID ? *pNextLevelID : 0L;

It's okay. pNextLevelID is checked for being non-NULL, and then either *pNextLevelID or 0L is assigned to dwNextLevelID.

-Erik

____________________________
The Godkiller - Chapter 1 available now on Steam. It's a DROD-like puzzle adventure game.
dev journals | twitch stream | youtube archive (NSFW)

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Thanks for the check.

I'm not really trying to understand the code atm, just trying to compile it

ErikH2000
Level: Legendary Smitemaster

Rank Points: 2796
Registered: 02-04-2003
IP: Logged

ross wrote:
There are LOTS and LOTS and LOTS of warnings. Mostly about friend declarations requiring a class key "friend X" should be "friend class X",

Okay, I just checked in fixes for these cases.

although there are fair few about dodgy typecasting (unsigned int x = -1) etc.

If you don't mind, just post the output of warnings, and we can decide how to deal with them. I like to stay warning-free.

Generally, we've tried to cast everything correctly and use the C++ static_cast and const_cast for most of it.

I don't intend to fix this immediately unless it stops compilation, but does anyone else see this on their compiler (non-gcc) ?

I'll fiddle with my VC settings later and see if I can get the warnings to come up.

-Erik


____________________________
The Godkiller - Chapter 1 available now on Steam. It's a DROD-like puzzle adventure game.
dev journals | twitch stream | youtube archive (NSFW)

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

The vast majority of warnings are those previously mentioned and the assignment of -1 to a uint. I'll try and find the specific instances where this occurs for you.

Also in the FileDialogWidget there are a lot of places where WCHAR are used on the assumption that HAS_UNICODE has been defined (when it hasn't). I'll send a patch for this when I can.

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Two warnings:

EditSelectScreen.cpp:128: warning: ISO C++ forbids declaration of `
X_CANCEL_BUTTON' with no type

There are several lines fairly close to this one....

and
CFileDialogWidget::SetDirectory(const WCHAR*)':
FileDialogWidget.cpp:457: warning: invalid conversion from `const WCHAR*' to `
short unsigned int*'

Should it be unsigned short ?

[Edited by ross on 05-18-2003 at 09:34 PM]

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

I'll just post all of the warnings in this single message for you

GameScreen.cpp:1261: warning: initialization of negative value `-1' to `UINT'
GameScreen.cpp:1261: warning: argument of negative value `-1' to `unsigned int'
GameScreen.cpp:1266: warning: choosing `CAttachableWrapper<T>::operator T&()
[with T = UINT]' over `CAttachableWrapper<T>::operator T() const [with T =
UINT]'
GameScreen.cpp:1266: warning: for conversion from `CAttachableWrapper<UINT>'
to `unsigned int'
GameScreen.cpp:1266: warning: because conversion sequence for the argument is
better
GameScreen.cpp:1273: warning: choosing `CAttachableWrapper<T>::operator T&()
[with T = UINT]' over `CAttachableWrapper<T>::operator T() const [with T =
UINT]'
GameScreen.cpp:1273: warning: for conversion from `CAttachableWrapper<UINT>'
to `unsigned int'
GameScreen.cpp:1273: warning: because conversion sequence for the argument is
better

LevelStartScreen.cpp:33: warning: ISO C++ forbids declaration of `
CX_LEFTRIGHT_MARGIN' with no type
g++ -c -g -I/Users/ross/devel/include/ -I/Users/ross/devel/include/SDL/ -I../DRODLib/ -c ListBoxWidget.cpp -o ListBoxWidget.o
ListBoxWidget.cpp: In member function `void
CListBoxWidget:ragPosButton(int)':
ListBoxWidget.cpp:187: warning: initialization to `const int' from `double'
ListBoxWidget.cpp:187: warning: argument to `int' from `double'
ListBoxWidget.cpp: In member function `void CListBoxWidget::CalcAreas()':
ListBoxWidget.cpp:833: warning: assignment to `Uint16' from `double'
ListBoxWidget.cpp:833: warning: argument to `short unsigned int' from `double'
ListBoxWidget.cpp:838: warning: assignment to `Sint16' from `double'
ListBoxWidget.cpp:838: warning: argument to `short int' from `double'

I'm feeling guilty now but you did ask

RestoreScreen.cpp:103: warning: ISO C++ forbids declaration of `
X_RESTORE_BUTTON' with no type
and lines following.


RoomWidget.cpp:116: warning: argument of negative value `-1' to `unsigned int'
and lines around there...

SettingsScreen.cpp:141: warning: ISO C++ forbids declaration of `CX_HELP_BUTTON ' with no type
and following lines.

should I stop for now?



[Edited by ross on 05-18-2003 at 09:44 PM]

Schik
Level: Legendary Smitemaster

Rank Points: 5453
Registered: 02-04-2003
IP: Logged

ross wrote:
Also in the FileDialogWidget there are a lot of places where WCHAR are used on the assumption that HAS_UNICODE has been defined (when it hasn't). I'll send a patch for this when I can.

I haven't looked at it in a while, but just so you know - IIRC all the paths, etc. are kept in Unicode strings and only converted to ASCII when actually used (in a non HAS_UNICODE build). So basically the only code that needs to be in #ifdef HAS_UNICODE blocks is code that opens files, gets directory listings, etc.

If you find some place where (i.e.) a file is opened with a Unicode filename that's not enclosed in the #ifdefs, then yeah, it's a bug.


____________________________
The greatness of a nation and its moral progress can be judged by the way it treats its animals.
--Mahatma Gandhi

ErikH2000
Level: Legendary Smitemaster

Rank Points: 2796
Registered: 02-04-2003
IP: Logged

ross wrote:
GameScreen.cpp:1266: warning: choosing `CAttachableWrapper<T>::operator T&()
[with T = UINT]' over `CAttachableWrapper<T>::operator T() const [with T =
UINT]'
GameScreen.cpp:1266: warning: for conversion from `CAttachableWrapper<UINT>'
to `unsigned int'
GameScreen.cpp:1266: warning: because conversion sequence for the argument is
better

I didn't really understand these. I checked in a more precise cast, which maybe doesn't help.

I'm feeling guilty now but you did ask

It's no big deal to fix them. For all the ones you've pointed out so far, I checked in fixes. If you don't mind, just post the warning output of any others you find, and I'll probably check in a fix. I don't care if it is a stupid-long list.

Unfortunately, VC doesn't show me the same warnings even with the warning level set to highest.

-Erik

____________________________
The Godkiller - Chapter 1 available now on Steam. It's a DROD-like puzzle adventure game.
dev journals | twitch stream | youtube archive (NSFW)

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

I've posted the warnings for GCC in another thread for your perusal. There is one place where I've cast blind (it's too late to look at the MK docs and work it out - must sleep).

Looks like I was too enthusiastic in thinking I would have this done in a day Usual over-optimistic programmer.

I have however got a libdrod.a sat in my lib directory waiting for me to get DROD compiled (YAY). I assume the majority of the problems in DROD itself will be problems with abs() again.

Also I still need to find the SetVar() and GetVar() for one of the DB classes to make sure the endian-ness is correct, I've forgotten where now, I'll check again tomorrow.

ErikH2000
Level: Legendary Smitemaster

Rank Points: 2796
Registered: 02-04-2003
IP: Logged

ross wrote:
I've posted the warnings for GCC in another thread for your perusal. There is one place where I've cast blind (it's too late to look at the MK docs and work it out - must sleep).

Looks like I was too enthusiastic in thinking I would have this done in a day Usual over-optimistic programmer.

Heh. Yeah, it always seems like it will take much less time.

I'm going to be much slower to respond to stuff for the rest of the week. I tend to work on DROD just Sunday and Monday.

Don't lose your momentum!

-Erik


____________________________
The Godkiller - Chapter 1 available now on Steam. It's a DROD-like puzzle adventure game.
dev journals | twitch stream | youtube archive (NSFW)

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

So now DROD compiles (as well as DRODLib) and to paraphrase some famous person or other "It's all over bar the linking".

No doubt there will be some Mac specific bugs (like endian-ness issues that the pre-proc warned about, but I couldn't find), I'll squash these as and when.

[Edited by ross on 05-20-2003 at 06:45 PM]

Schik
Level: Legendary Smitemaster

Rank Points: 5453
Registered: 02-04-2003
IP: Logged

ErikH2000 wrote:

ross wrote:
GameScreen.cpp:1266: warning: choosing `CAttachableWrapper<T>::operator T&()
[with T = UINT]' over `CAttachableWrapper<T>::operator T() const [with T =
UINT]'
GameScreen.cpp:1266: warning: for conversion from `CAttachableWrapper<UINT>'
to `unsigned int'
GameScreen.cpp:1266: warning: because conversion sequence for the argument is
better

I didn't really understand these. I checked in a more precise cast, which maybe doesn't help.

I wonder if this might be fixed by changing CAttachableWrapper to look like this:

template <typename T>
class CAttachableWrapper : public CAttachableObject
{
public:
	CAttachableWrapper(T data) : data(data) { }
	virtual ~CAttachableWrapper() { }
	T data;
	operator T&() {
		return data;
	}
	operator const T&() const {
		return data;
	}
};

At the very least, Visual Studio doesn't mind it. Not that it minded it the other way either...


____________________________
The greatness of a nation and its moral progress can be judged by the way it treats its animals.
--Mahatma Gandhi

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Wouldn't worry for now.

I got a few problems elsewhere ...I must link against Objc for SDL on MacOS but there is a definition of realloc in fmod AND in libobjc, which is stopping me linking. G++ wants to ignore -u and welcome to linker hell.

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Linker hell has got the better of me, I'm taking a break. I'll have another look at this Thursday ...

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Just got a message on the FMod forum that the __realloc problem has been fixed in a different release for Mac.

Hopefully I will get a little further tonight.

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

The Makefile for SGI does not include DrodStream in $(SOURCES)

Schik
Level: Legendary Smitemaster

Rank Points: 5453
Registered: 02-04-2003
IP: Logged

I wouldn't doubt if there are other omissions as well. I ported it to the SGI, then since my SGI compiler seems to have spotty at best Unicode support, I brought it back to the PC side to add Unicode support in. DrodStream was needed for Unicode files.

After the next release I'll take a look at getting it all working again on the SGI.


____________________________
The greatness of a nation and its moral progress can be judged by the way it treats its animals.
--Mahatma Gandhi

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

This is where I am at ... spot the absolutely idiotic mistake I made in my copy of you makefile.

ld: can't create output file: ../drod (Is a directory, errno = 21)

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

So .. soo... sooo close. I got this after a few seconds (Fatal signal: Bus Error (SDL Parachute Deployed)
)... but it was soo close. Time to visit the debugger methinks.



The slight image distortion is the screen capture not the app.

Schik
Level: Legendary Smitemaster

Rank Points: 5453
Registered: 02-04-2003
IP: Logged

First off, congrats!

Second, I can probably save you some time. This is very likely an endianness problem, unless you changed the code...

It's probably crashing when trying to load a level to run the demo.

grep for LittleToBig in DRODLib - some are commented out (and should be removed), but the ones that aren't should all be enclosed in #ifdef __sgi ... #endif

Your Mac should be big endian too, so it needs the same conversion as the SGI.

Good Luck!


____________________________
The greatness of a nation and its moral progress can be judged by the way it treats its animals.
--Mahatma Gandhi

ross
Level: Delver
Rank Points: 30
Registered: 05-17-2003
IP: Logged

Thanks for the tip. There are other problems as well (also no doubt endian) where all message text comes out as lil control codes, and you can't click on the buttons you have to click drag ... etc.. etc..

Schik
Level: Legendary Smitemaster

Rank Points: 5453
Registered: 02-04-2003
IP: Logged

The text being wonky is consistent with the endianness problem (I saw it all on the SGI...), as the text is Unicode, so the bytes are swapped.

The button clicking problem is new to me though.

____________________________
The greatness of a nation and its moral progress can be judged by the way it treats its animals.
--Mahatma Gandhi