java/5365: gcjh -stubs does not include correct files

java/5365: gcjh -stubs does not include correct files

[nallen]

>Number:         5365
>Category:       java
>Synopsis:       gcjh -stubs does not include correct files
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Jan 12 12:46:00 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Nicholas Allen
>Release:        3.0.3
>Organization:
>Environment:
Linux
>Description:
When gcjh -stubs is used the stubs file generates the wrong include statements. For example, if you use gcjh -stubs on java.lang.Object it generates a java/lang/Object.cc and a java/lang/Object.h file which includes the header file using #include <java.lang.Object.h>. The correct include should be #include <java/lang/Object.h>. It also does not include the gcj/cni.h file.

The stub functions also have JvFail in them and I can't find any references to this function in gcj/cni.h. 
>How-To-Repeat:
gcj -stubs java.lang.Object
>Fix:
It should generate #include <java/lang/Object.h>
>Release-Note:
>Audit-Trail:
>Unformatted:

Re: java/5365: gcjh -stubs does not include correct files

[tromey]

Synopsis: gcjh -stubs does not include correct files

Responsible-Changed-From-To: unassigned->tromey
Responsible-Changed-By: tromey
Responsible-Changed-When: Wed Jan 16 13:19:12 2002
Responsible-Changed-Why:
    I'm handling this.
State-Changed-From-To: open->closed
State-Changed-By: tromey
State-Changed-When: Wed Jan 16 13:19:12 2002
State-Changed-Why:
    I've checked in the fix.  It will appear in gcj 3.1.
    Thanks for your report.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5365

Re: java/5365: gcjh -stubs does not include correct files

[Tom Tromey]

The following reply was made to PR java/5365; it has been noted by GNATS.

From: Tom Tromey <tromey@redhat.com>
To: nicholas allen <nicholas.allen@ireland.sun.com>
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: java/5365: gcjh -stubs does not include correct files
Date: 16 Jan 2002 14:26:46 -0700

 >>>>> "Nicholas" == nicholas allen <nicholas.allen@ireland.sun.com> writes:
 
 Nicholas> I like the idea of thowing an exception (say
 Nicholas> java.lang.UnsupportedOperationException("Method not
 Nicholas> implemented yet")).  I know this would require include the
 Nicholas> header file for the exception but this can always be deleted
 Nicholas> and causes no harm if left in.
 
 I agree.  I'm making this change.
 
 Nicholas> It would also be nice if gcjh -stubs didn't generate a stub
 Nicholas> file if there were no native methods in the class file.
 
 This would be nice but it is hard to implement given the current
 code.  Feel free to submit another PR for this if you like.  Please
 make it low priority.
 
 Nicholas> Another idea would be too ask (or warn) the user if a file
 Nicholas> already exists before overwiting it with the stub file. This
 Nicholas> would prevent loosing implementations that have already been
 Nicholas> written by mistake.
 
 This would also be nice, and it worth another low priority PR... or a
 patch.
 
 Thanks,
 Tom

Re: java/5365: gcjh -stubs does not include correct files

[Tom Tromey]

The following reply was made to PR java/5365; it has been noted by GNATS.

From: Tom Tromey <tromey@redhat.com>
To: nallen@freenet.co.uk
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: java/5365: gcjh -stubs does not include correct files
Date: 15 Jan 2002 17:39:27 -0700

 >>>>> "Nicholas" == nallen  <nallen@freenet.co.uk> writes:
 
 Nicholas> When gcjh -stubs is used the stubs file generates the wrong
 Nicholas> include statements. For example, if you use gcjh -stubs on
 Nicholas> java.lang.Object it generates a java/lang/Object.cc and a
 Nicholas> java/lang/Object.h file which includes the header file using
 Nicholas> #include <java.lang.Object.h>. The correct include should be
 Nicholas> #include <java/lang/Object.h>. It also does not include the
 Nicholas> gcj/cni.h file.
 
 I agree this is a bug.  I've written a fix which I'll check in
 shortly.
 
 When I generate the stub file I do see an include for <gcj/cni.h>.
 I was unable to reproduce this part of the PR.
 
 Nicholas> The stub functions also have JvFail in them and I can't find
 Nicholas> any references to this function in gcj/cni.h.
 
 I agree this is less than perfect.  However, I don't think we want to
 export this JvFail.  We could make each stub throw a new exception,
 but that would mean including some exception header file (which in
 theory users would not actually be using).  What do you think?
 
 Tom