Python 3.8 Success

[tbpassin]

This is not an issue to report a problem, but to report success. I can see that the Master branch has made a lot of progress since I last installed rst2pdf using pip. I have found a few small issues and submitted pull requests on them.

Using my cloned code base with these fixes, I have been able to build a pdf document with Python 3.8 using a Sphinx workflow. The document contains 68 pages, and is not very complicated. It only has a TOC, a few internal links, some URL links, and some graphics. Simple or not, though, it builds. I have been building versions of this document for several years with Python 2.7, so I'm rally happy to get it working with 3.8. The computer system is Windows 10/Python 3.8/Sphinx 2.2/ ReportLab 3.5.32.

It doesn't look all that great, though, so I will have to learn more about styles.

[tbpassin]

I forgot to include one important thing. To build my document successfully, I also had to fix an issue with a module from jinja2. The file is loaders.py. I posted a fix for it at jinja issue 767 (pallets/jinja#767).

There could still be a problem with how template paths are presented to jinja2 (i.e., with the wrong path separators), but at least the module doesn't raise an unhandled exception any more.

[tbpassin]

I'm not going to say that the document build I succeeded with covers all code paths. There are still plenty of 2/3 issues on the Master branch. Simple ones I know about are:

replace *unicode* type with *str*
replace  *basestring* with *str*
import some modules using syntax from the absolute_import module from __future__.
change the import paths for urllib imports

So there's still work ahead. Also, I'm using Windows, and autotest doesn't run on it, so I don't have much test coverage.

[maphew]

Comment to add xref, the PR's are #824 and #826

[lornajane]

I'm not sure we need to keep this issue open but I love this success report :)

Also @tbpassin the tests now run with pytest, in case you want to try that out on Windows too