Repository navigation
Formatting fails when value-to-print is very large #607
Description
Reproduce with:
- Start bpython (happens in both
bpython version 0.15 on top of Python 3.5.1+ /usr/bin/python3andbpython version 0.15 on top of Python 2.7.11+ /usr/bin/python) - Do something that has at least two screenfuls of output when printed. Example:
["Hello"] * 1000
Expected result: boring stuff. Just the printed "value" of a list that contains a thousand "hello"s.
Actual result: The value, and then some formatting related characters that got garbled:
>>> ["Hello"] * 1000
['Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello',
'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello',
[... SNIP ...]
'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello',
'Hello', 'Hello', 'Hello', 'Hello', 'Hello', 'Hello']�[-26;1H
>>>
Note the trailing U+001B which smells like formatting gone wrong.
I'm not sure whether this is a bug in xfce4-terminal 0.6.3, bpython, bpython-curtsies, curtsies, or maybe something else entirely. However, bpython is the only program exhibiting this behavior, so I reported it here.
Activity
thomasballinger commentedon Apr 12, 2016
I bet this is some nonstandard terminal formatting that curtsies is doing. I can't reproduce at the moment, but will try soon with xfce.
thomasballinger commentedon Jun 6, 2016
/me still hasn't tried with xfce
I'm guessing something's happening to the math in Curtsies such that the "Cursor Home" command that moves the cursor
<ESC>[{ROW};{COLUMN}Hcame out with a negative row here. I checked and Blessings passes through negative numbers to its Terminal.move command, so this is possible.So probably one of the numbers in
cursor_pos[0] - offscreen_scrolls + self.top_usable_rowis bad, but I don't know which. I don't have a Linux machine around at the moment to check.thomasballinger commentedon Jun 7, 2016
It turns out this is happening in my terminal too, but iTerm2 swallows the invalid sequence! It's likely a Curtsies issue, these sequence may not be doing anything important, I'll look into it more.
thomasballinger commentedon Jun 13, 2016
I noticed in a gif I uploaded for another reason that the cursor disappears sometimes, and I think this may be related based on the Curtsies source: the sequence to make the cursor visible is written just after this sequence that is sometimes negative.
thomasballinger commentedon Aug 6, 2016
I suspect this is the problem but haven't nicely reproduced it. @BenWiederhake if it's not much trouble would you try
pip install git+https://github.com/thomasballinger/curtsies.git@fix-bpython-607to see if it fixes the problem?BenWiederhake commentedon Aug 6, 2016
Appears to work, thanks :D
(For fun I
pip uninstalled this version of curtsies again, as a sanity check of sorts, and indeed: with the old system version, it still shows the faulty behavior, so your fix definitely works.)thomasballinger commentedon Aug 7, 2016
Great, I've merged that and will do a Curtsies release before long, so at least fresh installations of Curtsies will have it fixed. I'll leave this open until I've done that Curtsies release.
sebastinas commentedon Oct 10, 2016
A new version of curtsies has been released in the meantime.