RFR: JDK-8243109: Bootcycle build failures after Nashorn removal

[Erik Joelsson]

On 2020-04-21 00:51, Magnus Ihse Bursie wrote:
> On 2020-04-21 00:29, Erik Joelsson wrote:
>> With Nashorn removed, the bootcycle build is now broken for builds 
>> that process markdown files for docs or man pages. Same goes for 
>> building with with a recent enough JDK 15 as your bootjdk. While a 
>> long term solution is being worked on, we need something to keep our 
>> CI builds from failing. For now that would simply be disabling these 
>> parts of the build if a jjs launcher cannot be found in the boot jdk.
>>
>> This patch checks the bootjdk for jjs. If not found, ENABLE_PANDOC is 
>> set to false. This will in turn prevent full docs from being enabled. 
>> For local bootcycle builds, I explicitly disable pandoc in 
>> bootcycle-spec.gmk.in. For the jib profile 
>> linux-x64-bootcycle-prebuilt, the "--enable-full-docs" configure arg 
>> is inherited from the main profile, so needed to neutralize that with 
>> an override.
>>
>> I had to reorder a little bit in configure.ac to have the bootjdk 
>> setup when processing JDK options.
>>
>> Webrev: http://cr.openjdk.java.net/~erikj/8243109/webrev.01/index.html
> Thanks for taking on this! LGTM.
>
Thanks!

Before pushing I realized I didn't quite finish the change where I 
introduced the JJS variable instead of using $(BOOT_JDK)/bin/jjs. Here 
is a new patch with this properly applied. This makes it possible to 
override JJS using a command line variable in configure and having it 
actually be used.

http://cr.openjdk.java.net/~erikj/8243109/webrev.02/index.html

/Erik

> For the long term solution, I still have a half-finished rewrite of 
> the javascript filters in Java locally. I didn't realize it could have 
> been needed prior to the Nashorn removal due to the bootcycle builds, 
> otherwise I would have prioritized this.
>
> /Magnus
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8243109
>>
>> /Erik
>>
>