Agree, but what you call "integration bugs" is 99% of the time, "old bugs in app code that are just uncovered by the new runtime". One example I always remember is: "SimpleTimeZone tz = (SimpleTimeZone)TimeZone.getTimeZone(...)" - this code, found in several places in projects where I worked by JDK 1.3's time, was broken by JDK 1.4; the cause is simple, developers who relied on the fact that getTimeZone() used to return a SimpleTimeZone - but this was never guaranteed by the API. Lucene's description of some bug related to ordering of reflection data is similar, the reflection APIs never made any promises of ordering and I am surprised that any professional Java developer who uses reflection in any significant extent ignores this fact. (This is sometimes one odd effect of relying too much in testing... you can have a test that succeeds through 10 years and 4 major JDK releases, just because some critical but non-documented runtime behavior was kept unchanged all that time. TDD is all nice and cool, but it doesn't replace actually knowing your stuff, or sometimes reading API documentation.)
Cisco's bug is more in a gray area, because JDK's behavior was changed (on Vista/Win7 only?) to prefer the IPv6 stack (due the the much- improved unified stack). But IPv6 support was available since 1.4 (1.5 on Windows); and we're not talking about some random program that happens to use network stack here and there, we're talking about a program which core work is networking, and everybody knew about the upcoming changes in JDK 7's networking, that was a release driver since the earliest plans. Now in all fairness, of all companies/groups, who most irritates me about JDK compatibility bugs is... Oracle; specifically, SQLDeveloper. I love SQLDeveloper, but it always craps out at any major JDK release. Now with 1.7 it's once again unusable - remarkably, incremental loading of rows when visualizing table data is all screwed up. So I have to keep JDK 1.6 only to run SQLDeveloper. It's irritating because that's a program from the same company that now provides JDK, so it looks like an ugly case of left hand not knowing what the right hand does. It's not a recent bug from the very latest JDK 7 builds; I noticed the problem months ago, so they just didn't test during the beta period, or didn't care. A+ Osvaldo On Aug 4, 8:44 am, Fabrizio Giudici <[email protected]> wrote: > On 08/04/2011 01:12 PM, opinali wrote:> That seems to be purely a bug in > Cisco's software, not supporting > > IPv6. This coming from Cisco, the company that's basically synonym > > for network gear... pretty odd. > > Just to be clear - that's a typical integration problem of a new > software stack. Java 7 is a new software stack and I expect lots of > integration problems. Including in my code: it already happened when I > moved to Java 5 from 1.4 and 6 from 5. It's pretty normal and it will be > a combination of bugs in Java 7 and the hosting enviroment, but in many > cases they won't be bugs, but things tried for the first time in a > different combination. Until I test my stuff with Java 7 I suppose it > won't work; I expect most of the tests to pass, but a non relevant need > of fixing and patching for the rest. > > This is very different from bugs in the core, like in HotSpot. > > -- > Fabrizio Giudici - Java Architect, Project Manager > Tidalwave s.a.s. - "We make Java work. Everywhere." > java.net/blog/fabriziogiudici -www.tidalwave.it/people > [email protected] -- You received this message because you are subscribed to the Google Groups "The Java Posse" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/javaposse?hl=en.
