Re: SISC 1.8.7

On Sun, Jan 25, 2004 at 05:34:53PM -0500, Isaac Morland wrote:
> 
> Anyhow, I have three bug-like reports (one of which actually is a bug).
> Just for your information, I'm running under Mac OS X 10.2, with Java
> 1.4.1
> 
> First, the Java system property caseSensitive doesn't work.  The Scheme
> (case-sensitive) works fine, it's just that the system property isn't
> read.  I guess most Schemers don't want case-sensitive symbols...
> 
> You can confirm the problem as follows:
> 
> setenv JAVAOPT '-Dsisc.caseSensitive=true'
> ./sisc
> (case-sensitive)
> 
> It prints #f.  Contrast with -Dsisc.characterSet=UTF-8 which works fine.
> 
> I was able to fix this by inserting
> 
>         this.caseSensitive =
>             ctx.getProperty("sisc.caseSensitive",
> defaultCaseSensitive).equals("true");
> 
> at line 69 of sisc/env/DynamicEnvironment.java.

Thanks, I've fixed this one in CVS.

> This causes compilation of sisc/modules/io/FileManipulation.java and
> sisc/modules/io/IO.java to fail.  It's easy enough to fix; in IO.java I
> replaced line 172 with:
> 
>                 try {
>                 	b.append(URLEncoder.encode(u.getRef(), "UTF-8"));
>                 } catch (UnsupportedEncodingException x) {
>                 	throw new Error ();
>                 }
> 
> and a couple of lines below:
> 
>                 try {
>                 	b.append(URLEncoder.encode(u.getQuery(),
> "UTF-8"));
>                 } catch (UnsupportedEncodingException x) {
>                 	throw new Error ();
>                 }
> 
> 
> And in FileManipulation.java I replaced line 67 with:
> 
>         String path;
>         try {
> 	        path=URLDecoder.decode(u.getPath(), "UTF-8");
> 	    } catch (UnsupportedEncodingException x) {
> 	    	throw new Error ();
> 	    }
> 
> Unfortunately, of course, this will no longer compile under Java versions
> before 1.4.  So it would be a real pain to fix for SISC.  Perhaps it could
> be solved similarly to Quantity.java by having a build parameter.

Perhaps.  We've known about the above problem for quite some time, but 
it has to run unmodified on Java 1.2/1.3 systems, which this just 
doesn't allow.  

> 
> Finally, I have a suggestion for the sisc.characterSet parameter: perhaps
> it should default to UTF-8?  Right now I believe it defaults to the Java
> platform default.  In the modern context it seems to me that using the
> universal encoding might make more sense than whatever parochial encoding
> happens to be set on the machine on which SISC is running.

Thats probably reasonable.  I'm forwarding this message to the 
SISC-devel mailing list in case anyone has other opinions on the default 
character set.

	Scott

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQFAFFKdr9IW4v3mHtQRAlM9AJ95MlOVi1Zw2WFKIC8usxQxn2SPJQCfZuwh
xK0LBU5jm8K9epu199C/mPQ=
=Nxyf
-----END PGP SIGNATURE-----