hotspot build problems

[Erik Trimble]

On 12/11/2010 1:38 AM, Volker Simonis wrote:
> On Sat, Dec 11, 2010 at 12:30 AM, Erik Trimble<erik.trimble at oracle.com>  wrote:
>> On 12/10/2010 1:24 PM, Dr Andrew John Hughes wrote:
>>>> Regarding the HPI-problem I only have a wild guess. Recently there was
>>>> a change which removed HPI (see:
>>>> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6348631). So
>>>> perhaps this interferes with your build JDK which is quite old as far
>>>> as I saw ("1.6.0-rc-b104"). Perhaps building with a newer Java 6
>>>> version or even Java 7 may solve the problem.
>>>>
>>> That doesn't seem to have been Per's problem (as he completed a build with
>>> selinux turned off) but your comment worries me a little.  Is there a plan
>>> to update OpenJDK6 so it will still be able to build 7?  Otherwise that's
>>> going to be a bootstrapping issue.  I expect, with 7 stabilising, we'll
>>> see more people wanting to build it over the coming months and having to
>>> build an old version of 7 first is going to make things overcomplicated.
>>>
>> We build internally with a JDK6 version as the bootstrap version, so I would
>> expect that any issue that pops up with building 7 using 6 would be fixed
>> quickly.
>>
>> That said, I *think*, but I'm not sure, that we're using 6u14 right now as
>> the bootstrap JDK6.  And, that's almost 2 years old at this point, so can't
>> see any issue with using any reasonably recent JDK6 or OpenJDK6 build as the
>> bootstrap.
>>
> Yes, but as far as I saw, Per used "1.6.0-rc-b104" which seems quite outdated...

I did see that, and I do believe that at least a portion of Per's 
problem arise from using such an antiquated version of JDK6.

I was responding to Andrew's concern that a functioning JDK 7 would be 
required to build JDK 7.  More specifically, that OpenJDK6 would 
continue to be able to build OpenJDK7.


-- 
Erik Trimble
Java System Support
Mailstop:  usca22-123
Phone:  x17195
Santa Clara, CA
Timezone: US/Pacific (GMT-0800)