| |||||||||||||
|
| |||||||||||||
Project Jigsaw (Mark Reinholds Blog)Project Jigsaw (Mark Reinholds Blog)Posted Dec 4, 2008 23:57 UTC (Thu) by allesfresser (subscriber, #216)Parent article: Project Jigsaw (Mark Reinholds Blog)
Yeah, good luck with that... I have one word in mind: "inertia". Reorganizing the entire Java platform when lots and lots of people already depend on it being a certain way? And many of those people being corporate developers with lots of legacy code? Hmmm...
(Log in to post comments)
Project Jigsaw (Mark Reinholds Blog) Posted Dec 5, 2008 0:27 UTC (Fri) by drag (subscriber, #31333) [Link]
Well those people can just keep running a older version of java, right?
Project Jigsaw (Mark Reinholds Blog) Posted Dec 6, 2008 5:35 UTC (Sat) by elanthis (subscriber, #6227) [Link]
No. Because then libraries end up depending on different versions, and there's no way to use two separate libraries that depend on two incompatible versions of a runtime.
Even in the Open Source world it can take larger projects and libraries to convert to newer runtimes and APIs. In the corporate world, it can take ages.
Breaking compatibility with any major, popular platform consumer just isn't a realistic option if you're trying to deliver a true end-to-end platform rather than just a prepackaged runtime+library kit.
Project Jigsaw (Mark Reinholds Blog) Posted Dec 6, 2008 16:59 UTC (Sat) by jengelh (subscriber, #33263) [Link]
Which is why we have symbol versioning... Glibc does it for ages already.
Project Jigsaw (Mark Reinholds Blog) Posted Dec 6, 2008 18:19 UTC (Sat) by nix (subscriber, #2304) [Link]
And it's modelled on the version from... Sun! So I suspect Sun can do
something similar in a modular JDK :)
Exactly. Posted Dec 5, 2008 14:36 UTC (Fri) by kirkengaard (subscriber, #15022) [Link]
The example of X.org going modular is valuable here. There is a reason that XFree86 did not go modular, or make the other significant changes that X.org has. That reason is inertia, as you say. Developers are used to a continuous development process with rules and aspects of the project that just work the way they are. Discontinuity is the only reason that inertia got broken -- otherwise, any similar effort would have been a minor fork, and likely unsuccessful.
It's the difference between evolution in a static environment vs. under the conditions of punctuated equilibrium. While Sun is in control of Java (I don't expect that to change anytime soon), development will most likely be done in natural progression from the way it is being done now. It would take a punctuation of that equilibrium to bring about this sort of change; its desirability alone will not overcome project inertia.
Exactly. Posted Dec 5, 2008 18:19 UTC (Fri) by dlang (✭ supporter ✭, #313) [Link]
are you using the xorg modularity as an example of modularity being good? or of it being bad?
I've seen people in the Xorg team comment that going modular did not provide the benifits that they were hoping for and in retrospect they should not have bothered.
Exactly. Posted Dec 18, 2008 10:52 UTC (Thu) by daenzer (✭ supporter ✭, #7050) [Link]
> I've seen people in the Xorg team comment that going modular did not
> provide the benifits that they were hoping for and in retrospect they > should not have bothered.
My impression is that's a minority opinion.
|
|
||||||||||||