Hi Andrew,
From my previous experience I can say that it's common practice
to indicate in commit messages who prepared and integrated
these particular bits and who reviewed them. Here is a couple
of examples where original fixes and backports are nearly identical,
but their commit messages refer to different persons:
Original fix: http://hg.openjdk.java.net/jdk8u/jdk8u-dev/jdk/rev/3502753a9d66
Backport: http://hg.openjdk.java.net/jdk7u/jdk7u-dev/jdk/rev/589c21e1aa30
Original fix: http://hg.openjdk.java.net/jdk8u/jdk8u-dev/jdk/rev/610da7dcd1be
Backport: http://hg.openjdk.java.net/jdk7u/jdk7u-dev/jdk/rev/26ba36d4400c
We treated this data rather in terms of responsibility - who is
responsible for preparing/verification/approving this particular
backport. And I agree with you that we should have used the original
OpenJDK bug ID in the commit messages, to be in line with another
'rule' in OpenJDK - to use the 'main' bug ID in comments for all
backports of the 'main' fix.
Surely, we are open for your suggestions on comments format - let's
find the version that makes everybody happy. :)
Regarding 6638712 and 6650759 - the backports were prepared from
scratch, taking
http://hg.openjdk.java.net/jdk7/jdk7/langtools/rev/22872b24d38c
http://hg.openjdk.java.net/jdk7/jdk7/langtools/rev/dda7e13f09fb
as the base. I discovered the fact that they are ported to IcedTea
during the investigation whether the compilation problem we faced
reproduces with other OpenJDK-based products.
Thanks,
Nikolay
On 25.06.2014 17:58, Andrew Hughes wrote:
----- Original Message -----
As this is a backport, there's no reason to allocate an OPENJDK6 bug
to it.
Will openjdk6 jcheck pass my changes with the 7-digit openjdk bug id? I
wasn't aware.
It allows either. :)
>The original information could have been used, perhaps with the
additional reviewer.
I am fine with any format. If you can point me to a web/wiki page with
proper guidelines - I would be happy to follow those as well.
>Now it looks like Azul wrote this patch, which I don't believe is the
case.
Unlike the previous change (VS2010 support) here I did not update
copyrights on files exactly because I do not see us as true authors for
this particular patch. So it was certainly not my intention to make this
patch look like it was authored at Azul. However I did want to give the
guys credit for doing proper testing of the produced bits as part of the
backport effort.
I've tended to just retain the original information e.g.:
changeset: 893:7aa071f95dac
user: prr
date: Wed Apr 30 13:10:39 2008 -0700
files: make/sun/font/FILES_c.gmk make/sun/font/Makefile
src/share/classes/sun/font/FileFontStrike.java
src/share/classes/sun/font/FontManager.java
src/share/classes/sun/font/TrueTypeFont.java
src/windows/classes/sun/awt/Win32GraphicsEnvironment.java
src/windows/native/sun/font/lcdglyph.c
test/java/awt/Graphics2D/DrawString/ScaledLCDTextMetrics.java
description:
6656651: Windows Look and Feel LCD glyph images have some differences from
native applications.
Reviewed-by: igor, tdv
but, as you say, that doesn't give credit for the backporting work. On the
other hand, if I'd
gone to the other extreme, over 90% of the OpenJDK 6 changesets would be
credited to me which
would look very odd!
The only guidelines I'm aware of are:
http://openjdk.java.net/guide/producingChangeset.html
which don't cover this situation. Maybe it's worth raising on the OpenJDK
discussion list?
Did you backport these changes from scratch? I was a little confused, because
the reference to the IcedTea backport patches made me think you might have used
those.
We will respond to the other mail regardimg Deepak's finding in a
separate email.
Thanks,
Ivan
*From:* Andrew Hughes <mailto:gnu.and...@redhat.com>
*Sent:* Tuesday, June 24, 2014 22:30
*To:* Ivan Krylov <mailto:i...@azulsystems.com>
*Cc:* jdk6-dev@openjdk.java.net <mailto:jdk6-dev@openjdk.java.net>
----- Original Message -----
Andrew, thank you.
Just pushed the changes.
BTW, icedtea patches 6650759-missing_inference.patch and
6638712-wildcard_types.patch are now redundant.
In future, if you're backporting changes, can you please keep the original
author and summary information?
From:
user: mcimadamore
6638712: Inference with wildcard types causes selection of
inapplicable method
Summary: Added global sanity check in order to make sure that return
type inference does not violate bounds constraints
Reviewed-by: jjg
to:
user: ikrylov
OPENJDK6-34: OpenJDK6-b31 backport of JDK-6638712 to openjdk6
Summary: Original bug synopsis-Inference of formal type parameter
(unused in formal parameters) is not performed
Reviewed-by: aph
Contributed-by: nikgor <niko...@azulsystems.com>
As this is a backport, there's no reason to allocate an OPENJDK6 bug
to it.
The original information could have been used, perhaps with the
additional reviewer.
Now it looks like Azul wrote this patch, which I don't believe is the
case.
Also, the following bugs may be closed OPENJDK6-32,33,34,35. I would
close those myself but I do not have the required permissions.
Done.
Thanks,
Ivan
On 23/06/2014 17:09, Andrew Haley wrote:
On 06/23/2014 02:05 PM, Ivan Krylov wrote:
The main motivation for this fix was exactly to fix building
JBoss EAP
certification bundle (for 6.2.0).
This fix exists in IcedTea (6650759-missing_inference.patch) but for
whatever reason was never promoted to openjdk6.
Ah, OK. That's fine, then.
Thanks,
Andrew.
--
Andrew :)
Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)
PGP Key: 248BDC07 (https://keys.indymedia.org/)
Fingerprint = EC5A 1F5E C0AD 1D15 8F1F 8F91 3B96 A578 248B DC07
