conformance-discuss

Download

conformance-discuss@openjdk.org

September 2012

4 participants
2 discussions

Re: [7u10] Request for approval: 7166055: Javadoc for WeakHashMap contains misleading advice
by Seán Coffey 16 Oct '12

16 Oct '12

I'd have to agree with allowing minor/simple javadoc updates also where specification changes are not implied. Even though Oracle mightn't always update their javadocs it shouldn't stop others from doing so (again for minor/simple/typo updates) I've run into arguments in past tough around what sort of javadoc updates do and do not imply spec. changes. Let's check with conformance team before deciding if this change is ok for an update release. Regards, Sean. On 18/09/2012 15:51, Andrew Hughes wrote: > > ----- Original Message ----- >> On 16/09/2012 1:26 AM, Phil Race wrote: >>> On 9/15/12 3:46 AM, David Holmes wrote: >>>> Phil, >>>> >>>> On 15/09/2012 2:57 AM, Phil Race wrote: >>>>> I really don't think its appropriate to push javadoc changes into >>>>> an >>>>> update release without >>>>> a really, really compelling reason that I don't see here. >>>> That is certainly true if they represent a specification change, >>>> but >>>> there is no semantic change here this is a simple clarification. >>> That would just rule it out completely. But we don't even >>> regenerate >>> javadoc for >>> the update releases and we have never randomly backported doc >>> comments, for >>> no obvious reason. So my reasoning and position stands. >> This is OpenJDK, it doesn't matter if "we" don't regenerate javadoc >> for >> update releases. And I have long thought that "we" should! I >> understand >> the issue with spec changes in update releases but I never understood >> a >> policy that would allow errors, misconceptions and mis-guidance to be >> set in stone instead of correcting them for the benefit of the user >> community. >> > +1 > > GNU/Linux distributions will make use of this new documentation in new builds, > even if the copies on the Oracle website aren't updated. The fact that you don't > want to jump through whatever hoops are needed to update your own copies should not > stop people from making minor updates (clarifications, typo fixes) at the OpenJDK level. > > I don't know how often jdk7u builds with docs are done at Oracle but there are currently > a number of warnings being thrown out by the build: > > ../../src/share/classes/java/awt/color/ICC_Profile.java:1069: warning - Tag @see: missing '#': "activateDeferredProfile()" > ../../src/share/classes/java/lang/invoke/MethodHandle.java:392: warning - Tag @link: reference not found: Objects.equals java.util.Objects#equals > ../../src/share/classes/java/util/Calendar.java:1717: warning - Tag @see: can't find setInternallySetState(int) in java.util.Calendar > ../../src/share/classes/java/util/Currency.java:685: warning - @throws tag has no arguments. > ../../src/share/classes/javax/swing/plaf/nimbus/NimbusStyle.java:854: warning - @return tag has no arguments. > ../../src/share/classes/javax/swing/plaf/nimbus/NimbusStyle.java:926: warning - @return tag has no arguments. > /home/andrew/builder/icedtea-jdk7/impsrc/javax/xml/bind/JAXBContext.java:262: warning - Tag @see: reference not found: S 7.4.1 "Named Packages" in Java Language Specification</a> > 7 warnings > > Are we supposed to retain these too? I can probably provide webrevs to fix these, but there's > no point if they aren't going to be accepted. > >> David >> >>> -phil. >>> >>>> David >>>> ------ >>>> >>>>> A reminder: Update releases aren't a free-for-all. You need to >>>>> exercise >>>>> judgement in what >>>>> has to go in and what is the case for it. We are up to 7u10 now. >>>>> We need >>>>> to be dialling >>>>> back the rate of change and focusing on JDK 8. >>>>> >>>>> -phil. >>>>> >>>>> >>>>> On 9/14/2012 12:56 AM, Shi Jun Zhang wrote: >>>>>> Hi all, >>>>>> >>>>>> I'd like to request for approval to push the following change >>>>>> into >>>>>> 7u10. >>>>>> >>>>>> Changeset in jdk8 >>>>>> http://hg.openjdk.java.net/jdk8/tl/jdk/rev/237e27c7ddc3 >>>>>> >>>>>> Webrev >>>>>> http://cr.openjdk.java.net/~zhangshj/jdk7u/7166055/webrev.00/ >>>>>> >>>>>> Reviewed by dholmes, mduigou >>>>>> >>>>>> Review thread >>>>>> http://mail.openjdk.java.net/pipermail/core-libs-dev/2012-May/010322.html >>>>>> >>>>>>

4 4

Re: [7u10] Request for approval: 7166055: Javadoc for WeakHashMap contains misleading advice
by Artem Ananiev 17 Sep '12

17 Sep '12

(Adding conformance-discuss to CC) On 9/17/2012 2:11 PM, Alan Bateman wrote: > On 17/09/2012 11:01, Artem Ananiev wrote: >> >> JavaDoc changes, even trivial corrections and clarifications, are >> potential compatibility issues. Previously unverifiable statement can >> become verifiable, and vice versa. We don't (can't!) change >> compatibility tests in update releases, therefore we can't accept any >> specification changes. > The discussion here relates to changes to non-normative text rather than > a specification change. I don't think it matters too much if such > changes are back-ported except that it might be a bit confusing to have > a mismatch between the source code and the docs on java.sun.com (or > docs.orcale.com nowadays). The difference between "non-normative text" and "specification" is often too subtle, and people may have different opinion what is what. I personally don't have any better ideas than to prohibit any spec changes in update releases, but I don't want to pretend to be a compatibility expert. Thanks, Artem > -Alan

2 1