RFR: JDK-8180281 --with-jtreg is broken for many use cases

[Erik Joelsson]

Looks good.

/Erik


On 2017-05-12 04:59, Magnus Ihse Bursie wrote:
> It turned out that run-test behaved badly when JT_HOME was not set. 
> Here's an updated version with a more helpful failure mode.
>
> http://cr.openjdk.java.net/~ihse/JDK-8180281-with-jtreg-is-broken/webrev.02 
>
>
> /Magnus
>
> On 2017-05-12 12:08, Magnus Ihse Bursie wrote:
>> JDK-8179537 was opened with a request to document the brokenness of 
>> --with-jtreg. Better to fix it than to document how it's broken.
>>
>> Now --with-jtreg works like this:
>>  * --without-jtreg will explicitly disable jtreg.
>>  * --with-jtreg=<jtreg home> will use the specified directory, and 
>> will do sanity checks that it is indeed a jtreg home pointed to, and 
>> not a binary.
>> * When no option is given, we try to locate jtreg using JT_HOME from 
>> the environment, and, failing that, by finding the jtreg executable 
>> in the path and deriving a JT_HOME from that. If not found, jtreg 
>> will be disabled.
>> * --with-jtreg (with no arguments) will try to locate jtreg 
>> automatically as described above, but if it fails, it will abort 
>> configure.
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8180281
>> WebRev: 
>> http://cr.openjdk.java.net/~ihse/JDK-8180281-with-jtreg-is-broken/webrev.01
>>
>> /Magnus
>