Request for Review: Java SE 8 Compact Profiles
kelly.ohair at oracle.com
Sat Dec 22 00:11:37 UTC 2012
On Dec 21, 2012, at 3:27 PM, David Holmes wrote:
> On 22/12/2012 1:18 AM, Kelly O'Hair wrote:
>> On Dec 20, 2012, at 10:18 PM, David Holmes wrote:
>>> top-level repo:http://cr.openjdk.java.net/~dholmes/8004265/webrev.top/
>> These comments in Main.gmk don't seem to make sense:
>> 117 # Note: This double-colon rule is intentional, to support
>> 118 # custom make file integration.
>> 119 images:: source-tips demos images-only
>> Do lines 117 and 118 just need to be deleted?
> No they are correct - sorry if they don't make sense. A "custom" makefile (such as the Oracle JDK closed makefile) may need to augment the images target (as we previously did in the old build). The :: rule allows for this custom images target to effectively concatenate it's recipe with the main one.
Never mond me, my eyes played tricks on me, I did not see :: just : :^(
>>> The main change is to simply add profiles and profiles-only as top
>>> level make targets (similar to images). There is also a change to
>>> remove the hardcoded version information (though this may be handled
>>> by a separate CR).
>>> jdk repo:http://cr.openjdk.java.net/~dholmes/8004265/webrev.jdk/
>> Can't cover the makefiles 100%, Erik would be best to look at some of
>> this, but this is what I have so far:
>> On JarReorder.java, it seems like you have just deleted a warning that
>> someone explicitly asked for
>> a class to be included, and also explicitly asked for that class to be
>> If we are changing the tool so that exclusion just silently trumps any
>> inclusion request, seems like we
>> should just do that and delete this message. I'm fine with that, but the
>> if(false) seems a bit terse.
> Yes ideally this change will trigger a closer look at jarreorder and how it is used. AFAIK those listings have been decaying. But the warning message was far too noisy for the profiles builds. I did not want to go down a path of trying to define per-profile reorder lists given that we haven't maintained this for the full JRE anyway.
Can we add a comment as to that being the reason for the if(false)? Maybe file a separate Issue to fix it someday,
or maybe toss the whole ball of JarReorder wax someday. ;^)
>> Why are some of the makefiles named with a ".txt" suffix? Like
> Because they aren't makefiles ;-) They are txt files that define named lists that happen to be compatible with makefile variable declarations.
But they aren't plain text files, right?
> These lists also get used by other tools eg javac and javadoc.
Do we have any convention for the file suffix on these yet? Or is the long term plan to just use .txt?
>> Overall, I have always been uncomfortable with these detailed
>> exclude/include lists when they get
>> down to listing specific class files, not that your changes are making
>> it any worse, but I do see this
>> as an opportunity to improve things in the long run by capturing the
>> specifics of our product shipments.
>> So no objections from me at this time, but at some point we need Erik to
>> check this out.
>> Unfortunately, everybody on build-infra will be busy for a few weeks
>> trying to get the cutover done. :^(
> Not to mention the Xmas/NewYear break. :(
Yeah, might be a limited vacation for some of us.
More information about the core-libs-dev