Re: [RELEASE] Python 3.9.0b1 is now available for testing

[Joseph Jenne via Python-list]

I used https://github.com/python/pyperformance pyperformance to 
compare Arch linux latest



Python 3.8.3 (default, May 17 2020, 18:15:42) [GCC 10.1.0] on linux
Type "help", "copyright", "credits" or "license" for more information.




against a vanilla build (configure make makeinstall) of python 3.9b1


Python 3.9.0b1 (default, May 19 2020, 21:09:14) [GCC 10.1.0] on linux
Type "help", "copyright", "credits" or "license" for more information.




I find all the bench marks seem to be slower in python 3.9b1.


38.json
===

Performance version: 1.0.1
Report on Linux-5.6.14-arch1-1-x86_64-with-glibc2.2.5
Number of logical CPUs: 4
Start date: 2020-05-31 04:00:24.503704
End date: 2020-05-31 04:22:44.961331

39.json
===

Performance version: 1.0.1
Report on Linux-5.6.14-arch1-1-x86_64-with-glibc2.31
Number of logical CPUs: 4
Start date: 2020-05-31 04:23:21.247268
End date: 2020-05-31 04:49:09.891889

### 2to3 ###
Mean +- std dev: 437 ms +- 5 ms -> 548 ms +- 7 ms: 1.25x slower
Significant (t=-96.22)

### chameleon ###
Mean +- std dev: 12.5 ms +- 0.1 ms -> 16.2 ms +- 0.2 ms: 1.30x slower
Significant (t=-111.53)

> ...

Is this because I haven't built in the same way as Arch or are there 
real slowdowns in this beta? Or even dumber have I got the results the 
wrong way round?

--
Robin Becker


Most builds of python included with distribution packages are built with 
various levels of optimization. I have experienced slowdowns from source 
built python of the same version as the distribution python even when 
using some optimization flags with the configure script. This appears to 
be normal behavior and is not cause for concern about the performance of 
python 3.9.0b1


--

Joseph Jenne


--
https://mail.python.org/mailman/listinfo/python-list

Re: Python-list Digest, Vol 201, Issue 9

[Joseph Jenne via Python-list]

On 2020-06-09 09:00, zljubi...@gmail.com wrote:


Well the problem that I am facing with is, that I have to establish interface 
between python and outer system.

Original question was about creation of input object (data that I have received from outer system). 
If I accept recommendation to use "from_" instead of "from", it could work, for 
processing input, because processing is under my control.

However, my process will create output object that I should json serialize and return back to outer system as a 
response to the input. If I will have "from_" object property instead of "from", I believe that I 
should write a custom object to json serializer in order to support changing names from "from_" to 
"from".


It should be possible to name it from_ and then insert it into the __dict__ as 
'from', although a custom serializer
would probably be preferable from a design standpoint.

--
https://mail.python.org/mailman/listinfo/python-list