RFR (S): JDK-8025933 Configure should support French cl.exe

[David Holmes]

On 8/10/2013 2:28 AM, Francis ANDRE wrote:
>
> Le 07/10/2013 11:32, Magnus Ihse Bursie a écrit :
>> On 2013-10-06 05:19, David Holmes wrote:
>>> Magnus,
>>>
>>> Echoing Naoto's comments I don't agree that we should be supporting
>>> this. In the original email a workaround was mentioned using an
>>> environment variable - if that is the case then I don't think we need
>>> to do anything here. Even if not, who knows how many other languages
>>> we might have to support in this way.
>>
>> I didn't think a simple patch would generate so much discussion. :-)
>>
>> I'm just about this close -->| |<-- to dropping this but let me just
>> elaborate on a few points here:
>> 1) The fix was indeed intended to be language independet; the comment
>> about Frech was unfortunate and should be rewritten.
>>
>> 2) The workaround mentioned was for JDK7. No such workaround is
>> possible for JDK8, and we will fail at configure time.
>>
>> 3) I think English should be the supported locale; unfortunately on
>> Windows platforms changing locale is system-wide and thus not possible
>> to change for only JDK build tools. Building in a virtual machine with
>> English locale to work around this is of course possible, but that
>> will make a slow build even slower.
>>
>> 4) At the very least, a simple fix that don't get configure
>> artificially caught when trying to check if the tools is okay, when it
>> actually is, would be okay.
>>
>>
>> New webrev fixing the comment:
>> http://cr.openjdk.java.net/~ihse/JDK-8025933-configure-support-french-cl.exe/webrev.02
>>
>>
>> If anyone still objects to this, let me known and I'll drop it.
> You must change in your patch the ' ' character following ersion by a
> dot '.' because in fact the character is X'FF' which is not displayable
> and replaced by a space.
>
> 00000000h: 43 6F 6D 70 69 6C 61 74 65 75 72 20 64 27 6F 70 ; Compilateur
> d'op
> 00000010h: 74 69 6D 69 73 61 74 69 6F 6E 20 4D 69 63 72 6F ; timisation
> Micro
> 00000020h: 73 6F 66 74 20 28 52 29 20 33 32 FF 62 69 74 73 ; soft (R)
> 32ÿbits
> 00000030h: 20 43 2F 43 2B 2B 20 76 65 72 73 69 6F 6E FF 31 ;  C/C++
> versionÿ1
> 00000040h: 36 2E 30 30 2E 33 30 33 31 39 2E 30 31 20 70 6F ;
> 6.00.30319.01 po
> 00000050h: 75 72 20 38 30 78 38 36 0A                      ; ur 80x86.
>
> By the way, I would prefer a fix like in the hotspot JDK7u with a user
> settable variable like ALT_MSVC_COMPILER_VERSION and
> ALT_MSVC_LINK_VERSION. If those variables are not defined, then use the
> make script. There are so many ALT_ variables to customize the build
> that it make sense to use ALT variables also in that case.

The new build in JDK has deprecated the use of ALT_ environment 
variables in favour of using configure options and variables. Hence 
Magnus's approach.

I just don't want to open the flood gates on language specific tool changes.

David
-----


> But you decide, it is just my pratical preference.
>
> Francis
>>
>> /Magnus
>>
>