[issue18726] json functions have too many positional parameters

Changes by Serhiy Storchaka : -- assignee: -> serhiy.storchaka resolution: -> fixed stage: patch review -> resolved status: open -> closed ___ Python tracker

[issue18726] json functions have too many positional parameters

Roundup Robot added the comment: New changeset db5fe5c4d09d by Serhiy Storchaka in branch 'default': Issue #18726: All optional parameters of the dump(), dumps(), https://hg.python.org/cpython/rev/db5fe5c4d09d -- nosy: +python-dev ___ Python tracker

[issue18726] json functions have too many positional parameters

Changes by Raymond Hettinger : -- assignee: rhettinger -> ___ Python tracker ___

[issue18726] json functions have too many positional parameters

Serhiy Storchaka added the comment: My remark was just about that I proposed to start a deprecation period at the time of developing Python 3.4. If this did accepted, the deprecation period would already finished and we would come up to the same code in 3.6. It doesn't matter now. --

[issue18726] json functions have too many positional parameters

Guido van Rossum added the comment: > Guido's decision on similar issue25628 is that changing keyword-or-positional > parameters to keyword-only is safe and can be made without deprecation. > If we started the deprecation period in 3.4, we could finish it in 3.6. That makes little sense. You

[issue18726] json functions have too many positional parameters

Serhiy Storchaka added the comment: Guido's decision on similar issue25628 is that changing keyword-or-positional parameters to keyword-only is safe and can be made without deprecation. If we started the deprecation period in 3.4, we could finish it in 3.6. Proposed simple patch changes all

[issue18726] json functions have too many positional parameters

Serhiy Storchaka added the comment: But what you will do with discrepancies between Python 2, Python 3 and simplejson? -- ___ Python tracker rep...@bugs.python.org http://bugs.python.org/issue18726 ___

[issue18726] json functions have too many positional parameters

Ezio Melotti added the comment: I have not seen any requests for this change. That's probably because either people have not realized that their code might be buggy, or because they have not realized that keyword-only args might be a solution to this problem (or maybe they did but didn't

[issue18726] json functions have too many positional parameters

Bob Ippolito added the comment: Other than when subclassing (which is actively discouraged), I haven't seen anyone try and use positional args for these APIs. I simply don't think this is an issue in practice. If you go far enough back in simplejson history, these module-global functions were

[issue18726] json functions have too many positional parameters

Serhiy Storchaka added the comment: If no one uses positional arguments then converting parameters to keyword-only will not break any code. -- ___ Python tracker rep...@bugs.python.org http://bugs.python.org/issue18726

[issue18726] json functions have too many positional parameters

Bob Ippolito added the comment: My evidence is only anecdotal. I haven't actively searched for code that uses json/simplejson to try and find cases that are using positional arguments. One way to test this assumption would be to release a version of simplejson that deprecates using positional

[issue18726] json functions have too many positional parameters

Raymond Hettinger added the comment: -1 Once the positional arguments are in the wild, you can't take them away without breaking a lot of code. This code was available as a third-party module prior to inclusion in Python and had many happy users. If the positional arguments had been a

[issue18726] json functions have too many positional parameters

Serhiy Storchaka added the comment: We don't want to create more obstacles to upgrading or converting from Python 2. But these obstacles already exists. The 10th parameter of dump() is encoding in Python 2 and simplejson, and default in Python 3. The 12th parameter of dump() is

[issue18726] json functions have too many positional parameters

Raymond Hettinger added the comment: Serhiy, please resist the urge to engage in API churn. We've had zero user requests to abandon the current API. I'm not a fan of the current positional arguments, but changing it isn't really worth it (creating disruption with nearly zero benefit, no new

[issue18726] json functions have too many positional parameters

Bob Ippolito added the comment: I tend to agree with Raymond here. While I also don't like the positional arguments, I don't think the benefit of API churn them is high enough to remove them. Other than this issue, I have not seen any requests for this change. I have seen problems because

[issue18726] json functions have too many positional parameters

New submission from Serhiy Storchaka: The json module functions have too many positional parameters: dump() -- 11 dumps() -- 10 load() -- 9 loads() -- 8 In most time only small part of these options is specified so users call these functions with keyword arguments for all parameters except

[issue18726] json functions have too many positional parameters

R. David Murray added the comment: This is not what we use keyword only arguments for. The standard practice in the stdlib is that arguments are arguments unless there is a good reason to make one keyword only. So I'm -1 on this proposal. -- nosy: +r.david.murray

[issue18726] json functions have too many positional parameters

Ezio Melotti added the comment: Serhiy has a point though, I wouldn't want to see something like json.dumps(someobj, True, False, True, False). -- ___ Python tracker rep...@bugs.python.org http://bugs.python.org/issue18726

[issue18726] json functions have too many positional parameters

R. David Murray added the comment: Ach. I didn't read carefully enough (not awake yet, I guess). Yes, boolean parameters are one of the things keyword only arguments are appropriate for. -- ___ Python tracker rep...@bugs.python.org

[issue18726] json functions have too many positional parameters

Changes by Arfrever Frehtes Taifersar Arahesis arfrever@gmail.com: -- nosy: +Arfrever ___ Python tracker rep...@bugs.python.org http://bugs.python.org/issue18726 ___