<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Andrew John Hughes wrote:
<blockquote
cite="mid:17c6771e0908191610w11b584b9xb53020fa87f58ba@mail.gmail.com"
type="cite">
<pre wrap="">2009/8/18 Andrew John Hughes <a class="moz-txt-link-rfc2396E" href="mailto:gnu_andrew@member.fsf.org"><gnu_andrew@member.fsf.org></a>:
</pre>
<blockquote type="cite">
<pre wrap="">2009/8/18 Jonathan Gibbons <a class="moz-txt-link-rfc2396E" href="mailto:Jonathan.Gibbons@sun.com"><Jonathan.Gibbons@sun.com></a>:
</pre>
<blockquote type="cite">
<pre wrap="">Andrew,
If this is a patch for jdk7, it does not appear to be a patch to a recent
copy
of 7.
</pre>
</blockquote>
<pre wrap="">It's against b69 which is the latest release (from Friday). The
patches are against the IcedTea forest so builds can be tested with
IcedTea as well.
Specifically, you do not seem to have the recent changeset to set
</pre>
<blockquote type="cite">
<pre wrap="">the source/target used to compile JDK to 7. [1]
</pre>
</blockquote>
<pre wrap="">Er... yes I do:
# Add the source level
LANGUAGE_VERSION = -source 7
JAVACFLAGS += $(LANGUAGE_VERSION)
# Add the class version we want
TARGET_CLASS_VERSION = 7
CLASS_VERSION = -target $(TARGET_CLASS_VERSION)
JAVACFLAGS += $(CLASS_VERSION)
JAVACFLAGS += -encoding ascii
JAVACFLAGS += -classpath $(BOOTDIR)/lib/tools.jar
JAVACFLAGS += $(OTHER_JAVACFLAGS)
but these only cover the rt classes and not the bootstrap classes.
</pre>
<blockquote type="cite">
<pre wrap="">While your patch does not directly conflict with any edits in that patch,
and
while the effect of your patch looks OK, in that patch I was extending the
precedent of TARGET_CLASS_VERSION to have an explicit macro for
(just) the version number, so that it is easy to change the value of (just)
the version number from the command line.
With that in mind, I would suggest something like the following for your
patch:
BOOT_SOURCE_LANGUAGE_VERSION = 6
BOOT_TARGET_CLASS_VERSION = 6
BOOT_JAVACFLAGS += -encoding ascii -source $(BOOT_SOURCE_LANGUAGE_VERSION)
-target $(BOOT_TARGET_CLASS_VERSION)
</pre>
</blockquote>
<pre wrap="">I didn't copy this for the 6 changes because I didn't immediately see
the point of using variables just for this single use. I forgot that
it is possible to override these from the command line, so I've update
the patch:
<a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~andrew/ecj/02/webrev.02/">http://cr.openjdk.java.net/~andrew/ecj/02/webrev.02/</a>
</pre>
<blockquote type="cite">
<pre wrap="">-- Jon
[1]
<a class="moz-txt-link-freetext" href="http://mail.openjdk.java.net/pipermail/compiler-dev/2009-July/001286.html">http://mail.openjdk.java.net/pipermail/compiler-dev/2009-July/001286.html</a>
On Aug 18, 2009, at 5:24 AM, Andrew John Hughes wrote:
</pre>
<blockquote type="cite">
<pre wrap="">Currently the javac calls for building the bootstrap tools (not the
classes for the final JDK, which correctly now use source and target
7) don't set an explicit source and target version.
The webrev:
<a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~andrew/ecj/02/webrev.01/">http://cr.openjdk.java.net/~andrew/ecj/02/webrev.01/</a>
sets these to 6 explicitly, as happens in the Ant builds performed by
langtools/jaxp/jaxws. This is noticeable especially when using ecj as
the bootstrap javac as it defaults to a version < 1.5, and the build
fails.
Ok to push?
Thanks,
--
Andrew :-)
Free Java Software Engineer
Red Hat, Inc. (<a class="moz-txt-link-freetext" href="http://www.redhat.com">http://www.redhat.com</a>)
Support Free Java!
Contribute to GNU Classpath and the OpenJDK
<a class="moz-txt-link-freetext" href="http://www.gnu.org/software/classpath">http://www.gnu.org/software/classpath</a>
<a class="moz-txt-link-freetext" href="http://openjdk.java.net">http://openjdk.java.net</a>
PGP Key: 94EFD9D8 (<a class="moz-txt-link-freetext" href="http://subkeys.pgp.net">http://subkeys.pgp.net</a>)
Fingerprint: F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
</pre>
</blockquote>
<pre wrap="">
</pre>
</blockquote>
<pre wrap="">
--
Andrew :-)
Free Java Software Engineer
Red Hat, Inc. (<a class="moz-txt-link-freetext" href="http://www.redhat.com">http://www.redhat.com</a>)
Support Free Java!
Contribute to GNU Classpath and the OpenJDK
<a class="moz-txt-link-freetext" href="http://www.gnu.org/software/classpath">http://www.gnu.org/software/classpath</a>
<a class="moz-txt-link-freetext" href="http://openjdk.java.net">http://openjdk.java.net</a>
PGP Key: 94EFD9D8 (<a class="moz-txt-link-freetext" href="http://subkeys.pgp.net">http://subkeys.pgp.net</a>)
Fingerprint: F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
</pre>
</blockquote>
<pre wrap=""><!---->
Is this version now ok? If so, I'll push it to the build gate using
the bug ID Kelly allocated for the same fix in JDK.
</pre>
</blockquote>
<br>
<br>
Andrew,<br>
<br>
I approve your webrev<span class="moz-txt-citetags"> </span><a
class="moz-txt-link-freetext"
href="http://cr.openjdk.java.net/%7Eandrew/ecj/02/webrev.02/">http://cr.openjdk.java.net/~andrew/ecj/02/webrev.02/</a><br>
My earlier confusion was caused by the fact that the corba Makefile is
not<br>
consistent with the jdk Makefile with respect to the use of
SOURCE_LANGUAGE_VERSION.<br>
It would be good to (separately) fix that inconsistency, but that does
not<br>
affect the validity of what you propose here.<br>
<br>
-- Jon<br>
</body>
</html>