From: Jeff A. <ja...@fa...> - 2019-03-08 22:58:18
|
Hi Albert: That's a very generous offer that I'm sure we'd like to take up if it proves possible for you to obtain funding. We should talk about the specific objectives as part of the application. Also, it's always interesting to hear what people are doing with Jython, where it provides a unique advantage, both your work and in Fiji. It looks like the PyPy "JVM backend" is equivalent to our org.python.compiler package, using jasmin as an external program where we use ASM. Jeff Jeff Allen On 08/03/2019 14:14, Albert Cardona wrote: > Hi Jeff, > > Thank you very much for your work on jython and for kindly taking the > time to answer my questions. > > I understand now that raising funds, which is within my means as an > academic, to support a full-time developer for 2 years would be > welcomed. Thank you very much for the encouragement. Realistically, it > will take over a year from now, if successful, to get started. While I > can make a convincing case for the use of jython within an academic > environment when tied to e.g. Fiji, the grant funding cycle is slow. > Finding a mature developer will also be hard, as within an academic > environment salaries are limited and generally non-competitive with > industry. But finding a junior or intermediate developer with > experience in java and python, willing to dive into a 2-year stint to > drive jython forward should be feasible. > > As an example of jython use in academia, I'd like to point here my > image processing tutorial: > https://www.ini.uzh.ch/~acardona/fiji-tutorial/ > > I have found teaching python to academics is far easier than any of > the many other alternatives. In my role as a professor in > neuroscience, I intend to introduce jython into biology courses for > senior undergraduates and for graduate students. The many libraries > for the JVM for image processing and numerical computing in general > are a great advantage; in particular, the extraordinary ImgLib2 > library https://github.com/imglib/ that abstracts data type, > dimensions and location without a performance loss. And the excellent > multithreading capabilities of the JVM, trivially accessible from > jython, are essential and a killer feature regarding the choice of > python+JVM over other options for modern processing of bioimagery, > ranging from many gigabytes to tens or hundreds of terabytes. To point > that modern scientific research in biology is very heavily dependent > on imaging and subsequent image processing. > > On RPython: I was perhaps misled by the jvm translator, a library that > I am not sure is or ever was part of the core RPython repository but > that can be found here > https://aecilius.cecs.anu.edu.au/mu/mu-client-pypy/tree/37ed6f826b4cdc1d8f8e732c22e24d380c762da9/rpython/translator/jvm > . > > Will be testing 2.7.2 alpha and report back, hoping in this way to > support jython development in the meanwhile, in addition to continue > to write tutorials based on jython. > > Best wishes, > > Albert > -- > Group Leader, HHMI Janelia Research Campus, USA. > Reader (Associate Professor), Dept. Physiol. Dev. Neurosci., > University of Cambridge, UK. > https://albert.rierol.net > https://www.janelia.org/lab/cardona-lab/ > https://www.pdn.cam.ac.uk/directory/albert-cardona > > Missatge de Jeff Allen <ja...@fa...> del dia dv., 8 de març > 2019 a les 9:18: >> Hi Albert, thanks for your interest. >> >> 1. I'm an active developer, and a few others, either submitting code (https://hg.python.org/jython/) or answering on issues or the web site. It is fair to say momentum is low, but 2.7.2b looks within reach and I'd really like to get the fix you mention out there. (I was on the verge of re-assessing our "blockers".) When that is out for evaluation, by good folk like you, I intend to revivify Jython 3. >> >> 2. We are a Python Software Foundation project and grateful for the continuing support we get in terms of infrastructure. No money changes hands AFAIK: I'm not sure we would know what to do with actual funding. (The PSF might.) We are all volunteers: I am working entirely in my own time. Contributors in the past sometimes had employer support/time. New contributors would be a help, and we try to be supportive when someone drops in. >> >> 3. I don't properly understand your question 3. (So simple answer "No and er ... ".) I understand RPython to be a restricted dialect underlying the PyPy implementation, but have not delved further. What do you mean by "JVM backend" here? We make Java bytecode, but it depends on the Jython runtime. I believe there was a compiler to Java source once, but support stopped before I joined the project. I have a lot of alternative ideas for the Jython 3 runtime (we can do much better), but I also see value in continuity. >> >> Jeff Allen >> >> On 07/03/2019 14:39, Albert Cardona wrote: >> >> Hi all, >> >> I would like to see jython development moving forward, and I would >> appreciate information on how to assist the existing jython developers >> to do so. >> >> First towards completing the 2.7.2 release towards out of alpha (it >> has a critical bug fix that I care a lot about: >> https://bugs.jython.org/issue2487). >> >> Second towards entering python 3 with a stable release (I've only been >> able to find https://github.com/jython/jython3 which is labels itself >> as an early alpha stage). >> >> Question 1: are any developers active presently? >> >> Question 2: where do you get your funding from, and would you commit >> to reach the above two milestones provided more funding was available, >> or should I aim at recruiting new developers? >> >> Question 3: rpython (from pypy) used to have a jvm backend in its git >> tree. I assume jython developers have considered pursuing that route, >> and if not, why not? >> >> Thanks very much. >> >> Albert Cardona >> -- >> Group Leader, HHMI Janelia Research Campus, USA. >> Reader (Associate Professor), Dept. Physiol. Dev. Neurosci., >> University of Cambridge, UK. >> https://albert.rierol.net >> https://www.janelia.org/lab/cardona-lab/ >> https://www.pdn.cam.ac.uk/directory/albert-cardona >> >> >> _______________________________________________ >> Jython-dev mailing list >> Jyt...@li... >> https://lists.sourceforge.net/lists/listinfo/jython-dev >> |