From: Bryan W. <bry...@po...> - 2004-09-25 05:54:03
|
brian zimmer wrote: > I'm interested in learning more about the group's interest in the > future direction of Jython. What are the primary areas of development > people wish addressed? Is it more important to work well with Java or > be feature-for-feature compliant with CPython? Do people like > jythonc? Should it be improved in any way? What about missing > modules from CPython such as select, sets or datetime? Should these > be implemented? Does anyone have significant changes to Jython they > have found useful? > > I'd like to know how people feel, what features are used the most and > what should be done for Jython. I'm not suggesting that I or anyone > currently on the development team will be able to do anything with the > results of this informal survey, but we might. > > Any information would be valuable. > > thanks, > > brian > hi brian, i'm currently using jython at work. we have a huge java project and my part is being done in jython. i develop my code and test everything with .py files. then i use jythonc to create a jar file which gets added to all the other jar files. i'm able to seamlessly call other team member's java code and they can seemlessly call mine. nobody can tell i'm using jython. so for me, jythonc is a _very_ important part. without it, jython would be a non-starter for me. jythonc was a bit annoying at first because i had to make it "disappear" into our build system so other developers could compile my jython code without "knowing" they were compiling jython. basically i eliminated the jythonc.bat file and created an ant task that calls the java on the jython compiler and separated out the files that were just for the compile which gets automatically synced as part of the make system. it would have been great if the jython website had an example of embedding jython in ant. another very annoying thing is that jython uses it's own deprecated methods. this requires me to remove all the deprecated warnings from the jython source code and compile my own version. this took care of the 3 deprecated methods, but i'm still getting other deprecated warnings when compiling with jdk 1.4.2. that i can't prevent even when compiling with the -deprecation flag. anyways, my point here is jythonc is very important to me, and you _must_ always remember that some of us use these tools in production build scripts and require it to run in ant. another improvement i see is in the exception stack trace if possible. when an exception occurs, it's sometimes very difficult to interpret it. sometimes i wish the 'million lines of jython internal methods" in the stack trace didn't show up... it's just freaks out not jython programmers to see that especially when it's deep and java calls jython which calls java which calls jython, etc. do you have any idea how many lines of jython stack trace you get that have nothing directory to do with your script? . is there an option to turn that off? the java programmers in my company see that and are instantly put off and won't look at jython. to answer your second part... jython can import and call any java package/object/method and visa-versa, so how can it work better with java :) it's _way_ more important to me for jython to be feature-for-feature compliant with CPython. it would be so cool to be able to take a cpython script and be able to run it jython. right now, all but the simplest scripts can do this with having to be tweaked or parts rewritten. bryan |