Open apiddington opened 4 years ago
Strange. Are the results actually the same everywhere?
I'm not sure what you mean by the 'same everywhere', but there is certainly a 5 fold time difference between v0.6.3 and v0.7.3 in exactly the same environment. I suppose I should do some more testing. I will try python2 on the virtual machine and compare.
I mean the actual plots, do they look the same? Is it the same number of steps required?
I hadn't looked carefully, but no they are all a bit different. That is v0.6.1 with python2, v0.6.1 with python3, v0.6.3 python3, and v0.7.3 and python3. Not very different and all quite acceptable.
I have tried with python2, python3 with adjustText 0.6.1 and 0.7.3 on the same virtual environment on Ubuntu 18.04, and while there is a little increase in runtime it's not very significant. I'll see what happens when I update my desktop to Ubuntu 20.04.
I wasn't able to install all the required python2 modules on Ubuntu 20.04. Hence my upgrade to python3.
Hopefully it won't be the 20secs from the virtual environment.
Thanks for the response. FYI here are my timings:
Python version: 2.7.17, adjustText version: 0.6.1, matplotlib version: 2.2.5 adjustText time: 2.9 secs adjustText time: 1.3 secs adjustText time: 0.0 secs Total adjustText time: 4.2 secs
Python version: 3.6.9, adjustText version: 0.6.1, matplotlib version: 3.3.1 adjustText time: 3.5 secs adjustText time: 1.4 secs adjustText time: 0.1 secs Total adjustText time: 4.9 secs
Python version: 2.7.17, adjustText version: 0.7.3, matplotlib version: 2.2.5 adjustText time: 1.9 secs adjustText time: 3.1 secs adjustText time: 0.1 secs Total adjustText time: 5.1 secs
Python version: 3.6.9, adjustText version: 0.7.3, matplotlib version: 3.3.1 adjustText time: 2.7 secs adjustText time: 3.5 secs adjustText time: 0.1 secs Total adjustText time: 6.2 secs
So basically the difference is not that dramatic after all?
I wouldn't bother with python2, it's basically not supported.
Hi, I have adjustText 0.6.1 installed using python2 on ubuntu 18.04. Running it with some 60 labels the response is almost instantaneous. Version 0.7.3 on a virtual machine running python3 on ubuntu 20.04 the same plot takes 20.5 secs. On the same machine and v0.6.3 it is 3.9 secs. Version 0.8b2 is a similar time to 0.7.3. I'm wondering why there is such a large increase in run time. Regards