public inbox for gcc-prs@sourceware.org
help / color / mirror / Atom feed
* java/1450: -M generates incorrect dependancies for .class files
@ 2000-12-20 12:27 Mike, Moreton
0 siblings, 0 replies; only message in thread
From: Mike, Moreton @ 2000-12-20 12:27 UTC (permalink / raw)
To: java-gnats
>Number: 1450
>Category: java
>Synopsis: -M generates incorrect dependancies for .class files
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: tromey
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Dec 20 12:20:04 PST 2000
>Closed-Date: Sun Jun 18 23:42:35 PDT 2000
>Last-Modified: Sun Jun 18 23:50:00 PDT 2000
>Originator: Mike Moreton
>Release: gcc version 2.95.1 19990816 (release)
>Organization:
>Environment:
>Description:
If a java file is dependant on a class that has a .class
file, then the -M and -MM options generate a dependancy to
the .java file rather than the .class file.
This was obviously deliberate, but it can mess up make, as
a class's ".class" file can change without the .java file
changing. (e.g. if this class is a sub-class of another
class, and that class changes) If this happens, make won't
spot the change, and hence the original java file won't get
recompiled, and your link will fail.
>How-To-Repeat:
You need three java files:
A.java:
public class A {}
B.java:
public class B extends A {}
C.java:
public class C {
public someMethod() {
B myB = new B();
}
}
run gcj -C B.java, then gcj -M C.java and you'll get a
dependancy on B.java not B.class
If you then change the name of a method in A that is called
from C, B.class will get recompiled, but because C.o does
not depend on B.class, it won't get recompiled.
>Fix:
In jcf-io.c function open_class, replace the lines
if (dep_name != NULL)
jcf_dependency_add_file (dep_name, 0);
with
jcf_dependency_add_file (filename, 0);
>Release-Note:
>Audit-Trail:
Formerly PR gcj/70
Responsible-Changed-From-To: apbianco->tromey
Responsible-Changed-By: tromey
Responsible-Changed-When: Thu Nov 4 21:37:58 1999
Responsible-Changed-Why:
The dependency-tracking code is mine, more or less.
State-Changed-From-To: open->feedback
State-Changed-By: tromey
State-Changed-When: Thu Nov 4 21:37:58 1999
State-Changed-Why:
I don't fully understand your example.
I followed your example. Then I tried compiling C.java:
creche. gcj -M C.java
C.o : ./A.java ./B.java /x1/egcs/install/share/libgcj.zip C.java
I think the dependency on A.java is telling. C depends on
A because the compilation of C causes B to be read,
which in turn causes A to be read.
So I don't see how your case could occur.
I don't doubt there is a problem here. It's just that
I don't understand what it is, and unfortunately your
example doesn't seem to show it. Can you change your
example to show the problem?
From: tromey@cygnus.com
To: Mike@sourceware.cygnus.com, Moreton@sourceware.cygnus.com,
apbianco@cygnus.com, java-gnats@sourceware.cygnus.com, tromey@cygnus.com
Cc:
Subject: Re: gcj/70
Date: 5 Nov 1999 05:37:58 -0000
Synopsis: -M generates incorrect dependancies for .class files
Responsible-Changed-From-To: apbianco->tromey
Responsible-Changed-By: tromey
Responsible-Changed-When: Thu Nov 4 21:37:58 1999
Responsible-Changed-Why:
The dependency-tracking code is mine, more or less.
State-Changed-From-To: open->feedback
State-Changed-By: tromey
State-Changed-When: Thu Nov 4 21:37:58 1999
State-Changed-Why:
I don't fully understand your example.
I followed your example. Then I tried compiling C.java:
creche. gcj -M C.java
C.o : ./A.java ./B.java /x1/egcs/install/share/libgcj.zip C.java
I think the dependency on A.java is telling. C depends on
A because the compilation of C causes B to be read,
which in turn causes A to be read.
So I don't see how your case could occur.
I don't doubt there is a problem here. It's just that
I don't understand what it is, and unfortunately your
example doesn't seem to show it. Can you change your
example to show the problem?
http://sourceware.cygnus.com/cgi-bin/gnatsweb.pl?cmd=view&database=java&pr=70
State-Changed-From-To: feedback->closed
State-Changed-By: bryce
State-Changed-When: Sun Jun 18 23:42:35 2000
State-Changed-Why:
- This is an ancient PR,
- we havn't heard back from the reporter,
- and I agree with Tom that there isn't a problem here:
$ gcj -C B.java
$ rm B.java
$ gcj -M C.java
C.o : C.java /usr/local/gcc/share/libgcj.zip ./B.class ./A.java
Changes to B's superclass, A, will still cause a rebuild of C
because gcj -M also reports the dependency on A.java.
From: bryce@albatross.co.nz
To: Moreton@sourceware.cygnus.com, Mike@sourceware.cygnus.com,
java-gnats@sourceware.cygnus.com, tromey@cygnus.com
Cc:
Subject: Re: gcj/70
Date: 19 Jun 2000 06:42:35 -0000
Synopsis: -M generates incorrect dependancies for .class files
State-Changed-From-To: feedback->closed
State-Changed-By: bryce
State-Changed-When: Sun Jun 18 23:42:35 2000
State-Changed-Why:
- This is an ancient PR,
- we havn't heard back from the reporter,
- and I agree with Tom that there isn't a problem here:
$ gcj -C B.java
$ rm B.java
$ gcj -M C.java
C.o : C.java /usr/local/gcc/share/libgcj.zip ./B.class ./A.java
Changes to B's superclass, A, will still cause a rebuild of C
because gcj -M also reports the dependency on A.java.
http://sourceware.cygnus.com/cgi-bin/gnatsweb.pl?cmd=view&pr=70&database=java
>Unformatted:
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2000-12-20 12:27 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2000-12-20 12:27 java/1450: -M generates incorrect dependancies for .class files Mike, Moreton
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).