If you find yourself constantly having to ask ‘did I change something in one of the jar projects? am I not seeing my changes because I forgot to do a clean?‘ etc., your environment sucks. Your process is crap. How good would a musician be who was constantly thinking ‘am I going to get an e flat if I hit these two keys?‘ or more like ‘funny, last time I hit those keys, I got an e flat, this time a d.. huh, wait, maybe I‘m flat because I have to spend some more time on my instrument, then I can resume my performance…‘
Per my last post, m2 has been dreamy. The best thing about it is that you just don't have to think about such crap anymore. I can just fly through the codebase, touching classes in the domain project, components, util, and web, and just hit the Apple-Option-P and the war publishes. This is thanks to the new m2 and also the fact that we are now using JBoss instead of Tomcat. Not just JBoss, but actually running the portal server inside eclipse. The only thing that sucks about this configuration is that you have to publish, for everything. Change a line in a JSP file, publish, a single directive in a CSS file, gotta publish, but still, publishing is not so bad! Restarting the whole portal server was taking 45s on my 4GB MacbookPro, but publishing only takes 15s!
I tried to fold JavaRebel into this, but no glory there. Right after wiring it into WTP, eclipse crashed, and then I was doing changes on classes, and it was not seeing it. I am going to try a few more things, but I think I am headed in the other direction I talked about before: I pulled my 8 core MacPro down into my front office and it starts in 33s and publishes in like 12s. But it has 8GBs of RAM. So I am thinking again about a RAM disk. The publication process is two big write operations, first in assembling the war in the target dir, and then copying it into the deploy dir, where it is then exploded by the app server. Thinking putting the workspace and the portal deploy dir on the RAM disk will take this down to 3 or 4s (after all, RAM is still something like 2000x faster than disk).. Even if this last ascent doesn‘t bring us to the couple second mecca, this is the best cycle we‘ve gotten to in eclipse by miles (also because the biggest joke about trying to get there with Tomcat is that while it will dynamically reload on certain changes, after a couple such routines, the whole server goes down in flames with a permgen error; JBoss has no such problem, and has halted, for any reason, maybe 2x in the past few weeks).
