You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Project Detroit - adding CPython to the Java VM #417
Sounds like they aim to rewrite JEP https://github.com/ninia/jep. I recommend to avoid giving too much on promises. We can take a look when Oracle delivers.
Back in 2015, there were great promises for GraalPython, they called it TrufflePython back then. They wanted to compile Python code, Java code, C from native extensions (via TruffleC, later via sulong) all into seamlessly compatible GraalVM bytecode. Modern GraalPython is an interpreter again, basically a faster CPython. It does not even compile Python code to Java bytecode. IBM had similar ideas btw, but the project never produced anything publicly available (AFAIK, correct me if I'm wrong).
Luckily, Jython has no paying custumers it can loose or has to satisfy. In that sense, there is no competition. They do their thing (or at least announce so) and we do ours.
In the article you linked, however, it is interesting that they finally recognize it's swimming against the stream (which is also Jython's problem), right now when one would expect AI could provide the heavy lifting for this problem.
It's a lot of work to accurately re-implement CPython, so I agree that from a JVM perspective it makes sence to just embed CPython inside the JVM. Project Detroit is an active OpenJDK project. We can track its progress here:
@ohumbel , JPython yes, brings back positive memories of the early days of Python in Java. I wrote a fairly complex database analysis and schema upgrade tool in JPython.
@ohumbel , JPython yes, brings back positive memories of the early days of Python in Java. I wrote a fairly complex database analysis and schema upgrade tool in JPython.
Do you remember why we changed the name? It always seemed more rational to me. (And does the PSF still own it?)
It's a lot of work to accurately re-implement CPython
Tell me about it! If they wanted to do it properly, they (their LLM) could read our code.
Activity
Stewori commentedon Mar 17, 2026
Sounds like they aim to rewrite JEP https://github.com/ninia/jep. I recommend to avoid giving too much on promises. We can take a look when Oracle delivers.
Stewori commentedon Mar 17, 2026
Back in 2015, there were great promises for GraalPython, they called it TrufflePython back then. They wanted to compile Python code, Java code, C from native extensions (via TruffleC, later via sulong) all into seamlessly compatible GraalVM bytecode. Modern GraalPython is an interpreter again, basically a faster CPython. It does not even compile Python code to Java bytecode. IBM had similar ideas btw, but the project never produced anything publicly available (AFAIK, correct me if I'm wrong).
Luckily, Jython has no paying custumers it can loose or has to satisfy. In that sense, there is no competition. They do their thing (or at least announce so) and we do ours.
In the article you linked, however, it is interesting that they finally recognize it's swimming against the stream (which is also Jython's problem), right now when one would expect AI could provide the heavy lifting for this problem.
wfouche commentedon Mar 17, 2026
It's a lot of work to accurately re-implement CPython, so I agree that from a JVM perspective it makes sence to just embed CPython inside the JVM. Project Detroit is an active OpenJDK project. We can track its progress here:
ohumbel commentedon Mar 18, 2026
The naming of the .sh File is interesting: https://github.com/openjdk/detroit-python/blob/master/jpython.sh
Reminds me of our very beginnings...
wfouche commentedon Mar 18, 2026
@ohumbel , JPython yes, brings back positive memories of the early days of Python in Java. I wrote a fairly complex database analysis and schema upgrade tool in JPython.
jeff5 commentedon Mar 31, 2026
Do you remember why we changed the name? It always seemed more rational to me. (And does the PSF still own it?)
Tell me about it! If they wanted to do it properly, they (their LLM) could read our code.
wfouche commentedon Apr 6, 2026
The name changed with version 2.0. JPython 1.x became Jython 2.0
I really do not know.