Frank Dana

This is technically an API change. In fact, it's a fairly significant one _if_ you happen to be abusing the API with things like `pydot.Graph("name", None, "graph", True, True, False)`,...

I really am wondering if I should make `d.write('path', 'format')` legal, though. Or, alternatively, if we should try to infer the format based on the filename passed. (e.g.): ```python d.write('/tmp/mygraph.svg')...

Ideally I'd get un-lazy and add some tests before this gets merged, anyway. So, no rush.

In addition to the tests I've so far still been too lazy to write, I wonder if there's enough old, breakable code out there to make it worth providing a...

> But this is one of the easiest migrations I've seen: > > ```diff > - import pydot > + import pydot.legacy as pydot > ``` _If_ it works. I...

Wait... I'm an _idiot_. Nothing about `dot_parser` has to change for `pydot.legacy` — the same extraction of `obj_dict`s and discarding of objects that makes subclassing a pain will actually _save_...

I almost have the subclassing PR (#495) finished with tests that create objects from `pydot.core` subclasses and verify that they don't mutate, but since this PR is really dependent on...

I mean, if we're going to do that, we should probably display the device ID as well, and might even consider showing the list of supported packet types the device...

I also don't think it would be the worst UX if we turned the "landing page" of each device's preferences into a status display (rather than a settings section), and...

@uncharted-user Windows support, in particular, is _not_ very well tested... but I was able to pair a Windows 10 system running KDE Connect with GSConnect on my Fedora desktop, relatively...