Benson already looks at the Survey and posted some of his thoughts. Here are my thoughts based on the questions:
1) The interesting thing about this question is that there are a LOT of relatively new users of CXF. For an "established" project, that is kind of exciting. Over 70% of the users have used it for less than 2 years with half of them under one year. That's a lot of new users and definitely shows we need to help them out getting started. Since the first release of CXF was in July of 2007, I wonder how people have used it for more than 5 years. :-) 2-5) Were targetted at figuring out how CXF is used in businesses. To me, the interesting thing is that nothing really stands out as a stronghold. It's used in a wide variety of industries and for a wide range of solutions. (although integration does stand out a bit) This is good and bad. Good that it's genericly useful for a wide audience. Bad in that it's hard to really target any one area for improvement. 6-7) Like 2-5, these questions show that CXF is used by ALL sizes of companies. The usage is fairly evenly spread out over the entire company size selection. Again, hard to target specific things. 8-9) These were a bit surprising to me, but may be a reflection of the results of 2-7. These questions show that there are a low number of "CXF applications" per business, but each application hosts a number of CXF endpoints. I may try digging into the raw results to see how these correlate to the company size metrics. 10) Ommitting Jetty was a serious mistake. 12 write in votes for Jetty shows it IS used, but Tomcat is deinitly number 1 by far. We definitely need more tests and docs for tomcat. 11) No real surprises here (although missing OSX was dumb on my part). Linux is #1 followed by Windows. 12) Again, no surprises. Eclipse #1 as expected. I guess my only real surprise was the lack of more Emacs/VI write ins. They must finally be giving up after all these years. ;-) 13) No real surprise here either. The main projects listed are the same ones we collaborate with to make sure things work seemlessly. 14) People like Spring. Nice to know. :-) 15) No real surprises. Kind of wish I would have separated Axis1 and Axis2, but nothing major. 16) OK. We need better docs, tutorials, and documented examples. I guess I should have expected that. :-) 17) Looks like we as a community do a good job supporting our users. That's a good thing. 18-19) I liked these questions. JAX-WS and JAX-RS are #1 and #2 which REALLY justifies the work we've done with the JAX-WS and JAX-RS implementations and getting the TCK's passing and such. Security being #3 (#1 on 19) also justifies the work there, but also shows we need to work harder on improving it. Some of that work involves working closer with WSS4J and such to get bugs fixed there. I know Dave and Glen have started filing bugs and patches there which is great. I was slightly surprised the coloc stuff wasn't rated higher, but I guess CXF really targets distributed stuff, not in-VM stuff as much. My only other major thought: we need more JMX instrumentation. 20) All of these specs are "neutral", with a slight lean toward possitive. This is probably another question where delving into the raw data may be interesting to see if large companies want these more than the smaller companies or similar. 21) Kind of expected. The new databindings are only slightly interesting. The other spec things are more so. 22) OK. We have too many dependencies. Not sure what can be done other than documenting better what deps are required and when. I was surprised that the OSGi stuff wasn't a bit higher. 23) Yea, users want a little of everything. Want fries with that as well? 24) Not my area. I know very little about these so I really don't have a comment. 25) I definitely encourage everyone to read all the comments. A lot of good comments that really helped boost my spirits. Several good suggestions and such as well. -- Daniel Kulp [email protected] http://dankulp.com/blog
