On Fri, 4 Mar 2022 23:33:46 GMT, Magnus Ihse Bursie <i...@openjdk.org> wrote:

>> When doing reproducible builds, controlling the build time is imperative. To 
>> make this as easy as possible, some changes are needed in the build system.
>> 
>>  * If source-date is set to anything other than "updated" (meaning that it 
>> should be determined at build time), then the default value of 
>> --with-hotspot-build-time should be the same time.
>> 
>> * If the industry standard SOURCE_DATE_EPOCH is set when running configure, 
>> the default value for --with-source-date should be picked up from this 
>> variable. (If the command line option is given, it will override the 
>> environment variable however.)
>> 
>> * Finally the code can be made a bit clearer that we can set and export 
>> SOURCE_DATE_EPOCH to SOURCE_DATE in spec.gmk, unless we're using the 
>> "updated" (determined at build time) value.
>
> Magnus Ihse Bursie has updated the pull request incrementally with one 
> additional commit since the last revision:
> 
>   Update build html file

As for RPM builds (by whom?) suddenly being build with 
--enable-reproducible-build, I would not think it is a matter of concern. For 
linux, the only real difference --enable-reproducible-build does is add the 
--date flag to jmod (and to the bootjdk jar, if it supports it), and runs a 
special tool on src.zip to make it reproducible. The only real reason this flag 
exists (and not just is the default behavior) is out of an abundance of 
caution. On Windows, it triggers the `-experimental:deterministic` flag, which 
has the scary sounding word "experimental" in it. But we have been running with 
this option enabled for all Oracle Windows builds for years by now, I think, 
without seeing any problems.

So I'm actually contemplating removing the --enable-reproducible-build flag, 
and just make that behavior default. There's very little reason not to have 
this turned on always, and many good reasons for having it.

-------------

PR: https://git.openjdk.java.net/jdk/pull/7660

Reply via email to