Feature or enhancement
Proposal:
In #117500, we create a nice mechanism to register and retrieve the source code of an arbitrary code object. It was a bug fix so we want to keep it as private as possible - it was for PyREPL usage only. However, we did put effort into the design so it can be used in a wider range in the future, and now is the future!
We just passed beta freeze so we have enough time to make it a new feature. The code and mechanism is really simple, it's just about documentation (and whether we want to merge some interfaces).
There are two orthogonal areas I want to bring into discussion, for each I have two proposals:
- How do we want the interface
- Keep it as it is, meaning the user needs to explicitly register and retrieve source code from a complete separate pair of interfaces.
- Combine
getlines and _getline_from_code, making code an optional argument to getlines - take the path when the optional code object is passed in.
- How public do we want it to be
- Make it public to all users.
- Make it public internally so at least
inspect and pdb can use it. (we don't document it, but we keep inspect.getsourcelines work)
In either combination, we have the full backwards compatibility - nothing will be broken. It's more about how much maintenance effort we want to put in and how easy we want the user to use the feature.
An obvious example usage is that, if you want to debug some interactive code, either through pdb.run(), or debug command in pdb. You don't have the source code symbols now, even though it's fully available. With this capability, we can make pdb.run() work nicely with string source code.
cc: @pablogsal
Has this already been discussed elsewhere?
No response given
Links to previous discussion of this feature:
#117500
Feature or enhancement
Proposal:
In #117500, we create a nice mechanism to register and retrieve the source code of an arbitrary code object. It was a bug fix so we want to keep it as private as possible - it was for PyREPL usage only. However, we did put effort into the design so it can be used in a wider range in the future, and now is the future!
We just passed beta freeze so we have enough time to make it a new feature. The code and mechanism is really simple, it's just about documentation (and whether we want to merge some interfaces).
There are two orthogonal areas I want to bring into discussion, for each I have two proposals:
getlinesand_getline_from_code, making code an optional argument togetlines- take the path when the optional code object is passed in.inspectandpdbcan use it. (we don't document it, but we keepinspect.getsourcelineswork)In either combination, we have the full backwards compatibility - nothing will be broken. It's more about how much maintenance effort we want to put in and how easy we want the user to use the feature.
An obvious example usage is that, if you want to debug some interactive code, either through
pdb.run(), ordebugcommand inpdb. You don't have the source code symbols now, even though it's fully available. With this capability, we can makepdb.run()work nicely with string source code.cc: @pablogsal
Has this already been discussed elsewhere?
No response given
Links to previous discussion of this feature:
#117500