[Scummvm-devel] [Scummvm-cvs-logs] SF.net SVN: scummvm: [23511] scummvm/trunk/engines/kyra/resource.cpp

Max Horn max at quendi.de
Thu Jul 20 23:16:30 CEST 2006


Am 20.07.2006 um 22:49 schrieb lordhoto at users.sourceforge.net:

> Revision: 23511
>           http://svn.sourceforge.net/scummvm/?rev=23511&view=rev
> Author:   lordhoto
> Date:     2006-07-15 13:30:36 -0700 (Sat, 15 Jul 2006)
>
> Log Message:
> -----------
> Use the FS Api to load all .PAK and .APK files. (I just made a few  
> quick tests, but before this should be integrated to branch-0-9-0  
> there should be a test for it on *every* platform, since I'm not  
> quite sure if FilesystemNode::displayName is useable for this...)
>
> Modified Paths:
> --------------
>     scummvm/trunk/engines/kyra/resource.cpp

[...]

> +	for (FSList::const_iterator file = fslist.begin(); file !=  
> fslist.end(); ++file) {
> +		if (file->displayName().hasSuffix("PAK") || file->displayName 
> ().hasSuffix("APK")) {
> +			assert(loadPakFile(file->displayName()));
> +		}
>  	}
>  }

[...]

No, that is not a valid use of displayName() (although it works now  
by chance). The display name really is there as a "human readable"  
name of the file, to be displayed to the user. It would be valid for  
me to add a "###" prefix to the result of displayName, and code must  
continue to work correctly.

You really should be using path() here instead.

Granted, though, FilesystemNode and File do not mesh quite perfectly  
right now (to put it nicely, I could find harsher words... *g*). I  
never got quite around completing the transition away from paths to  
FilesystemNode objects... ideally, there should be a File::open  
variant which takes a FilesystemNode object.


Cheers,
Max




More information about the Scummvm-devel mailing list