First - please edit subject so we stop getting Re: added ad-infinitum :)
Second - this is a general build issue relating to cygwin etc so I've
cc'd build-dev and bcc'd jdk8-dev.
I suspect your version of cygwin may be too new for the 8u build. You
should check out the build-dev archives over the last few weeks:
http://mail.openjdk.java.net/pipermail/build-dev/
David
On 5/05/2017 8:53 AM, Guy Bonneau wrote:
No success with the change to the profile.
Guy
Le 04/05/17 à 15:33, *Michal Vala * <mv...@redhat.com> a écrit :
On 05/04/2017 07:43 PM, Guy Bonneau wrote:
>Good point!
>
>
>This is what I have found. The file _the.jars.contents is a text file
that has a list of classes. There is many hundred of them.
>The command: /usr/bin/grep -e '\.class$'
/cygdrive/c/Projects/jdk-8/build/windows-x86_64-normal-server-fastdebug/images/lib/_the.jars.contents
yields nothing. It is empty and create an empty
_the.rt.jar.contents.tmp file with nothing. I checked and it is indeed
a file size of 0 bytes.
>
>
>However the command /usr/bin/grep -e '\.class^M$'
/cygdrive/c/Projects/jdk-8/build/windows-x86_64-normal-server-fastdebug/images/lib/_the.jars.contents
yields all the classes.
>
>
>I checked with a binary editor the file: _the.jars.contents and all
the line ending are Windows style. Not Linux style. It seems the tool
invoked that creates the _the.jars.contents see Windows as the OS and
put Windows style EndOfLine. Then Grep pattern fails because it
searches for Linux-style EndOfLine.
>
>
>Note that my JDK which might be invoked to run Java outside of the
Cygwin folder installation might create Windows style EndOfLine. This
could be the issue.
>
>
>Thus my guessing is the script must be modified with a grep pattern
that can automatically extract the lines ending with .class with
either Linux or Windows style EndOfLine.
>
>
>If you can give me a quick fix I could try it.
You may try to add this to your ~/.bash_profile in cygwin
export SHELLOPTS
set -o ignc
>
>
>Thanks
>Guy
>
>
>
>
>Le 04/05/17 à 10:04, Weijun Wang <weijun.w...@oracle.com> a écrit :
>>Looks like the exploded build succeeds and there is something wrong
with the image build.
>>
>>Can you try if the JDK inside
/cygdrive/c/Projects/jdk-8/build/windows-x86_64-normal-server-fastdebug/jdk
(i.e. the exploded build) works fine?
>>
>>The last few lines show
>>
>>/usr/bin/grep -e '\.class$'
/cygdrive/c/Projects/jdk-8/build/windows-x86_64-normal-server-fastdebug/images/lib/_the.jars.contents
>
/cygdrive/c/Projects/jdk-8/build/windows-x86_64-normal-server-fastdebug/images/lib/_the.rt.jar.contents.tmp
>>make[2]: *** [CreateJars.gmk:268:
/cygdrive/c/Projects/jdk-8/build/windows-x86_64-normal-server-fastdebug/images/lib/_the.rt.jar.contents]
Error 1
>>
>>Can you rerun that grep command and see if _the.rt.jar.contents.tmp
is empty?
>>
>>--Max
>>
>>On 05/04/2017 09:54 PM, Guy Bonneau wrote:
>>>I removed the JDK 8 and left the JDK 7 (1.7.0_80). No success same
error. I check the config log and saw some complaining about the JRE
7, So I removed this one as well and left only the JDK 7 to be found.
No success and same error at the same place.
>>>
>>>Guy
>>>
>>>Le 04/05/17, Guy Bonneau <guy.bonn...@videotron.ca> a écrit :
>>>>I have both oracle JDK 7 and JDK 8 installed on the default Java
installation folder of my computer (Need both of them while working
with Eclipse and miscellaneous project). I would have expected the
configuration script to automatically choose the JDK 7 rather than JDK
8 as a Bootstrap. I'll uninstall Oracle JDK 8 and restart the build
and let you know.
>>>>
>>>>Thanks
>>>>Guy
>>>>
>>>>Le 04/05/17, David Holmes <david.hol...@oracle.com> a écrit :
>>>>>On 4/05/2017 9:25 PM, Guy Bonneau wrote:
>>>>>>Did it. Both configuration and build logs are now shared at:
>>>>>>
>>>>>>https://www.dropbox.com/sh/6xbmjzxu8xlbsr0/AADBHa-Zn7owJpyeihlmej9pa?dl=0
>>>>>
>>
>>
--
Michal Vala
OpenJDK QE
Red Hat Czech
