Fwd: Re: RFR(xxs): 8072935: Fix missing newline at end of file after 8067447

David Holmes david.holmes at oracle.com
Thu Feb 12 23:42:29 UTC 2015


On 12/02/2015 11:27 PM, Volker Simonis wrote:
> Hi,
>
> I don't think we should enforce a minimum Version for GCC (actually we
> would have to enforce an exact version because the master builds at
> Oracle are only tested with exactly one compiler version - see
> https://wiki.openjdk.java.net/display/Build/Supported+build+platforms).
>
> Different users have different requirements and if somebody for some
> reason has to use an older/newer version of GCC we shouldn't make that
> unnecessarily hard. But issuing a warning during the configure step
> would be fine for me.

At some point as developers move our C++ code into the 21st century 
we're going to be implicitly requiring a minimum gcc version for things 
to work correctly. That minimum version already exists today we just 
don't know what it is. Being forced to work with ancient compilers is 
one of the things that has added to the historical baggage in hotspot 
code. :( Given how conservative we are with our official Oracle JDK 
compilers, and how the community always wants bleeding edge compiler 
support, I'm a little surprised at the resistance to this. :)

> I think that compiling a project with different compilers or different
> versions of a compiler is a very good overall health check. In this
> case we detected a file without new-line at the end of files. I think
> everybody agrees this is something we do not want to have, no
> difference if it is objected by the official build tools.
>
> I'd like to propose a new test in jcheck instead which checks for this
> situation.

I have no issue with that. Something normally warns me about this anyway 
... I think it might be webrev.

Cheers,
David

> Regards,
> Volker
>
>
> On Thu, Feb 12, 2015 at 1:49 PM, Erik Joelsson <erik.joelsson at oracle.com> wrote:
>> We don't have a minimum version currently. I agree that it would be a good
>> idea, but if we define one, wouldn't we also need to take responsibility of
>> testing that it keeps working?
>>
>> /Erik
>>
>>
>> On 2015-02-12 12:58, David Holmes wrote:
>>>
>>> Please see attached - do we have a minimum gcc version set in configure?
>>> If not, should we? I would think yes.
>>>
>>> David
>>
>>



More information about the build-dev mailing list