Cannot move only x or y coordinate in Screen session

[abstiles]

This affects move_x, move_y, and the location context manager (when only one of x or y is specified). The cause appears to be Screen's terminfo, which does not seem to define entries for "hpa" or "vpa". To test, I did the following:

$ tput -T screen cup > /dev/null; echo $? 
0
$ tput -T screen hpa > /dev/null; echo $? 
1
$ tput -T screen vpa > /dev/null; echo $? 
1

Note that changing the environment variable TERM to xterm fixes the issue.

It would be helpful if Blessings were able to internally workaround this problem transparently. As it stands, I cannot use these functions at all if I want my program to work under either Screen or Tmux (which uses Screen's terminfo).

[jquast]

I agree. I think this is only a problem for 'screen', so far as I can tell. They can be fixed by assigning an anonymous function for this case.

            if self._kind == 'screen' and not self.hpa:
                self.hpa = lambda col: u'\x1b[{}G'.format(col + 1)
            if self._kind == 'screen' and not self.vpa:
                self.vpa = lambda col: u'\x1b[{}d'.format(col + 1)

[jquast]

this is fixed as you recommend, "internally workaround this problem transparently", in a fork if you wish to need it, https://pypi.python.org/pypi/blessed

[abstiles]

Much appreciated, I'll check it out. Any plans to contribute any of your patches back here? I don't see any pull requests for fixes like this. Have they been explicitly declined by the Blessings maintainer?

[jquast]

I definitely tried to contribute all of my fixes and enhancements, beginning last year. The pull requests were mostly ignored/stalled -- Look in "closed" pull requests in this project and you can see them. He's more than welcome to have them all back in his own project, but I'm very sure he won't take the time to do so.

[erikrose]

I'm happy to take a patch to this effect. I certainly had a hand in stalling out those old PRs, but the sheer size of them (rather than being small, atomic changes) was also a factor. I would like to get a lot of that stuff upstreamed.

[jquast]

I'm afraid I can't do that. For example, this change cannot be tested without @as_subprocess -- as your first test case that tests xterm-256color causes all subsequent tests to be tested as xterm-256color.

[jquast]

This issue is resolved in 'blessed-integration' branch by a method of "proxying" terminal capabilities. https://github.com/erikrose/blessings/blob/blessed-integration/blessings/formatters.py#L144-148

[jquast]

This issue is resolved in blessed, which is API compatible with blessings, by a ParameterizingProxyString class, which is provided for missing terminal capabilities, from end of blessed.formatters.resolve_attribute():

and, for special terminals, such as 'screen', provide a Proxy
ParameterizingString for attributes they do not claim to support,
but actually do! (such as 'hpa' and 'vpa').