RFR: JDK-8136782: Introduce a build/configure wrapper

[Erik Joelsson]

On 2015-12-04 14:42, Erik Joelsson wrote:
>
>
> On 2015-12-04 12:50, Magnus Ihse Bursie wrote:
>
>>>>
>>>>> 144 *       // For certain dependencies where a legacy 
>>>>> distribution mechanism is
>>>>> 145  *       // already i place, the "javare" server layout is 
>>>>> also supported
>>>>>
>>>>> "in place". Also, missing period.
>>>>>
>>>>> 150  *       // buildnumber (optional), files and checksumfile is 
>>>>> possible for
>>>>>
>>>>> "build number", "checksum file"
>>>>>
>>>>> *       // aritfacts following the standard layout.
>>>>>
>>>>> "artifacts"
>>>>>
>>>>> *       // For other files, use checksumpath and paths instead
>>>>>
>>>>> "checksum path"
>>>>>
>>>>> 62  * input.build_unix_os
>>>>> 63  * input.build_unix_cpu
>>>>> 66  * input.build_unix_platform
>>>>>
>>>>> What is build_unix_cpu, etc? Sounds fully incomprehensible to me. :)
>>>> Clarified
>> I still think this is not so good. What does build_unix_os contain? 
>> "Windows" or "Cygwin"? Is it used only on Windows? And why "unix"? Is 
>> the whole reason for this to be able to separate 32 and 64 bit cygwin?
> Yes, and it would apply the same if we were to use msys. On a 64bit 
> windows, you can run either a 32bit or 64bit unix emulation layer. I 
> thought unix_emulation_layer was a bit long so shortened it to unix. I 
> agree the name is not good.
>> In the build system, we have a notion of "os env", which is the same 
>> as "os" in most cases, but isn"cygwin" or "msys" on Windows. Maybe we 
>> should use the same terminology here, if we describe the same thing.
> os_env only describes msys or cygwin, not the bitness. I need both to 
> be able to pick the correct executables.
>
Ok, I give in, here is a new version using "osenv" instead of "unix". I 
could not figure out a better combination of variable names than this.

http://cr.openjdk.java.net/~erikj/8136782/webrev.top.03/

/Erik