tag:blogger.com,1999:blog-6204005997345471829.post6677165943955431111..comments2023-03-25T11:24:24.318+00:00Comments on Bad Concurrency: Now Offically an OpenJDK ContributorMichael Barkerhttp://www.blogger.com/profile/16232688819921959379noreply@blogger.comBlogger10125tag:blogger.com,1999:blog-6204005997345471829.post-68148623722006456042011-08-01T22:04:09.038+01:002011-08-01T22:04:09.038+01:00I've still got quite a bit to do. It's ta...I've still got quite a bit to do. It's taken quite a while so far mostly 'cause I'm snatching weekends and evenings to work on it. I haven't really scratched the surface of the JDK/Hotspot, but I will at some point put together a blog on my experiences.Michael Barkerhttps://www.blogger.com/profile/16232688819921959379noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-30382052532884211922011-07-30T23:00:06.271+01:002011-07-30T23:00:06.271+01:00Wonderful!
Can you talk about how you went about ...Wonderful!<br /><br />Can you talk about how you went about navigating around the codebase and how you figured out how everything worked?<br />How long did it take, BTW?Mohamed Banahttps://www.blogger.com/profile/05666785617374419041noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-69440965708606669182011-07-25T19:11:34.759+01:002011-07-25T19:11:34.759+01:00@George There was a bit of interest from one of th...@George There was a bit of interest from one of the JRuby guys on the MLVM list, so there is some interest from some the of the language guys.<br /><br />You're right it could take a little while to push it through into Java.Michael Barkerhttps://www.blogger.com/profile/16232688819921959379noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-88884684266248389102011-07-25T19:06:06.374+01:002011-07-25T19:06:06.374+01:00@mikera Yes, although I have a few more steps to w...@mikera Yes, although I have a few more steps to work through before I get there. John Rose's proposal suggests a new byte code VRETURN that would support multiple return values.Michael Barkerhttps://www.blogger.com/profile/16232688819921959379noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-59381861558624026682011-07-25T18:53:10.424+01:002011-07-25T18:53:10.424+01:00Perhaps you should propose something like this the...Perhaps you should propose something like this the the Ceylon guys or other JVM language designers. Getting this into Java might take ages...George Moschovitishttps://www.blogger.com/profile/04736280934738427219noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-38186851966733579462011-07-24T22:58:20.677+01:002011-07-24T22:58:20.677+01:00Great that you are taking this on - it's been ...Great that you are taking this on - it's been my No.1 item on my wish list for the JVM for quite some time!<br /><br />Hopefully the same approach can be applied to efficiently allow multiple return values? I've wished for this many times when writing simulations or graphical code that returns 3D vectors..... having to allocate a new object or pass around a pre-allocated object to act as temporary storage is trult ugly....Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-64983686370200758942011-07-22T08:43:23.402+01:002011-07-22T08:43:23.402+01:00This is a great feature, I hope to see this in Jav...This is a great feature, I hope to see this in Java sooner, rather than later. Is there a place where we can vote for / support this?George Moschovitishttps://www.blogger.com/profile/04736280934738427219noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-14422847286290865142011-07-21T09:23:23.814+01:002011-07-21T09:23:23.814+01:00Good point and the same would follow for Java. E....Good point and the same would follow for Java. E.g. using a value type in a generic collection would required it to be boxed onto the heap.<br /><br />The post provides some excellent technical detail, however I disagree with his conclusion. The notion that the implementation doesn't matter grates against the notion of <a href="http://mechanical-sympathy.blogspot.com/2011/07/why-mechanical-sympathy.html" rel="nofollow">Mechanical Sympathy</a>, which is essential when building high performance software. I think I'll need to put together a separate post on that.Michael Barkerhttps://www.blogger.com/profile/16232688819921959379noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-400701164330093042011-07-20T22:50:48.154+01:002011-07-20T22:50:48.154+01:00C# structs are not always allocated on the stack. ...C# structs are not always allocated on the stack. Is actually much more complicated than that.<br /><br />http://blogs.msdn.com/b/ericlippert/archive/2010/09/30/the-truth-about-value-types.aspxgnashhttps://www.blogger.com/profile/12702643427008702523noreply@blogger.comtag:blogger.com,1999:blog-6204005997345471829.post-47257261271659116162011-07-20T22:13:59.467+01:002011-07-20T22:13:59.467+01:00Hi Michael,
Thanks for tackling this! As you say,...Hi Michael,<br /><br />Thanks for tackling this! As you say, it's a very useful feature for situations where maximum performance is needed. I'll be following it with interest.<br /><br />Best,<br />IsmaelIsmael Jumahttps://www.blogger.com/profile/17398483226873559286noreply@blogger.com