RE: 8+3-filesystem support

[Jeroen Frijters]

Dr. Torsten Rupp wrote:
> Attached to this email there is a patch. Please try it. It is 
> still not a nice implementation, because I still did not
> removed the backslash-tests to keep it compatible with the
> current implementation in Classpath.
> If everybody agree it would be a nice idea to separated the 
> pathname/filename handling functions into separated methods 
> and clean-up the code.

Let me start out by saying that I'm very much in favor of making the 8.3
support explicit, but I also have a couple of comments:

I would like to move the eightThreeFilesystem flag to VMFile (so I can
make it a constant).

Clearly the current approach of testing for '\\' is lame, but we
shouldn't replace it with something equally lame. As I see it there are
several issues:
- UNC paths (\\server\path)
- drive letter prefixes
- 8.3 length restrictions
- alternate separators (DOS/Windows support '/' as well as '\\')

I may have missed some. I think we should have a flag (in VMFile) for
each of these different issues. This makes the code clearer (IMHO) and
also makes it easy to disable particular functionality by setting a
particular flag to a constant false. That way, e.g. systems that don't
care about 8.3 can disable it and support for it will not be compiled
in.

Also, in your patch this looks incorrect:
+    if ((separatorChar!='\\') || eightThreeFilesystem)

Regards,
Jeroen