You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've built full-screen apps where of course I would have no need for patch_stdout(). More recently I started enriching "regular" scripts with occasional progress bars and prompts. For what I need a full screen app seemed kind of overkill, and I liked the idea for the experience to be easily able to gracefully degrade when the output is being piped into log files.
I wish the documentation around patch_stdout() would go a lot more in depth and perhaps enumerate all widgets that would and would not "bypass" the patched standard output, together with good and bad patterns. From certain examples, it seems like its context is only wrapping the use of certain widgets while other threads print to the terminal. Yet, I've run into situations where if I don't wrap a print in a patch_stdout() context, the print ruins the terminal output for a progress bar widget from a different coroutine than the one the print is coming from.
So, I've been lazily sprinkling with patch_stdout() all over the place, without thinking too much. I'd use it generously and a bit too naively, without trying to understand things too well and perhaps counting too much on the fact that its implementation looked like it should be re-entrant. Of course I ended up with a little bit of a nightmare at hand.
Maybe I should just be less stubborn and build yet another full-screen app instead.
In any case, here's some of the nightmarish stuff I have been able to cause thus far:
ValueError: write to closed file
ValueError('I/O operation on closed file.')
RuntimeError('File descriptor 0 is used by transport <ReadUnixTransport closed=False reading=True 0x7fc771440380>')
RuntimeError: File descriptor 0 is used by transport <ReadUnixTransport closed=False reading=True 0x7fd0b404c380>
Error in sys.excepthook:
object address : 0x7fc7626ef760
object refcount : 1
object type : 0x7fc776a2f640
object type name: ValueError
object repr : ValueError('I/O operation on closed file.')
lost sys.stderr
Original exception was:
object address : 0x7fc7627eba60
object refcount : 5
object type : 0x7fc776a30da0
object type name: RuntimeError
object repr : RuntimeError('File descriptor 0 is used by transport <ReadUnixTransport closed=False reading=True 0x7fc771440380>')
lost sys.stderr
File "setup.py", line 2173, in setup
confirmation = await session.prompt_async()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "venv312/lib/python3.12/site-packages/prompt_toolkit/shortcuts/prompt.py", line 1220, in prompt_async
return await self.app.run_async(
^^^^^^^^^^^^^^^^^^^^^^^^^
File "venv312/lib/python3.12/site-packages/prompt_toolkit/application/application.py", line 886, in run_async
return await _run_async(f)
^^^^^^^^^^^^^^^^^^^
File "venv312/lib/python3.12/site-packages/prompt_toolkit/application/application.py", line 734, in _run_async
with self.input.raw_mode(), self.input.attach(
File "/usr/lib/python3.12/contextlib.py", line 137, in __enter__
return next(self.gen)
^^^^^^^^^^^^^^
File "venv312/lib/python3.12/site-packages/prompt_toolkit/input/vt100.py", line 165, in _attached_input
loop.add_reader(fd, callback_wrapper)
File "uvloop/loop.pyx", line 2441, in uvloop.loop.Loop.add_reader
File "uvloop/loop.pyx", line 772, in uvloop.loop.Loop._add_reader
File "uvloop/loop.pyx", line 762, in uvloop.loop.Loop._ensure_fd_no_transport
RuntimeError: File descriptor 0 is used by transport <ReadUnixTransport closed=False reading=True 0x7fd0b404c380>
NoneType: None
File "/usr/lib/python3.12/contextlib.py", line 137, in __enter__
return next(self.gen)
^^^^^^^^^^^^^^
File "venv312/lib/python3.12/site-packages/prompt_toolkit/input/vt100.py", line 165, in _attached_input
loop.add_reader(fd, callback_wrapper)
File "uvloop/loop.pyx", line 2441, in uvloop.loop.Loop.add_reader
File "uvloop/loop.pyx", line 772, in uvloop.loop.Loop._add_reader
File "uvloop/loop.pyx", line 762, in uvloop.loop.Loop._ensure_fd_no_transport
RuntimeError: File descriptor 0 is used by transport <ReadUnixTransport closed=False reading=True 0x7fd0b404c380>
NoneType: None
File "venv312/lib/python3.12/site-packages/prompt_toolkit/shortcuts/utils.py", line 139, in render
renderer_print_formatted_text(
File "venv312/lib/python3.12/site-packages/prompt_toolkit/renderer.py", line 813, in print_formatted_text
output.flush()
File "venv312/lib/python3.12/site-packages/prompt_toolkit/output/vt100.py", line 693, in flush
flush_stdout(self.stdout, data)
File "venv312/lib/python3.12/site-packages/prompt_toolkit/output/flush_stdout.py", line 33, in flush_stdout
stdout.buffer.write(data.encode(stdout.encoding or "utf-8", "replace"))
ValueError: write to closed file
NoneType: None
File "setup.py", line 2475, in <module>
print()
File "venv312/lib/python3.12/site-packages/prompt_toolkit/shortcuts/utils.py", line 164, in print_formatted_text
render()
File "venv312/lib/python3.12/site-packages/prompt_toolkit/shortcuts/utils.py", line 139, in render
renderer_print_formatted_text(
File "venv312/lib/python3.12/site-packages/prompt_toolkit/renderer.py", line 813, in print_formatted_text
output.flush()
File "venv312/lib/python3.12/site-packages/prompt_toolkit/output/vt100.py", line 693, in flush
flush_stdout(self.stdout, data)
File "venv312/lib/python3.12/site-packages/prompt_toolkit/output/flush_stdout.py", line 33, in flush_stdout
stdout.buffer.write(data.encode(stdout.encoding or "utf-8", "replace"))
ValueError: write to closed file
NoneType: None
I understand that providing a minimal code snippet to reproduce these issues would be better, and perhaps I'll find the time to do so in the future. Meanwhile, is there any chance from the little info I could provide you'd have any hint as to how I should be doing instead and what I shouldn't be doing?
Thank you!
The text was updated successfully, but these errors were encountered:
I suppose that part of the issue may also be that the renderer height is unknown. Guessing this since the bottom toolbar for the progress bar is not showing up and I also get a:
WARNING: your terminal doesn't support cursor position requests (CPR).
I've built full-screen apps where of course I would have no need for patch_stdout(). More recently I started enriching "regular" scripts with occasional progress bars and prompts. For what I need a full screen app seemed kind of overkill, and I liked the idea for the experience to be easily able to gracefully degrade when the output is being piped into log files.
I wish the documentation around
patch_stdout()
would go a lot more in depth and perhaps enumerate all widgets that would and would not "bypass" the patched standard output, together with good and bad patterns. From certain examples, it seems like its context is only wrapping the use of certain widgets while other threads print to the terminal. Yet, I've run into situations where if I don't wrap a print in a patch_stdout() context, the print ruins the terminal output for a progress bar widget from a different coroutine than the one the print is coming from.So, I've been lazily sprinkling
with patch_stdout()
all over the place, without thinking too much. I'd use it generously and a bit too naively, without trying to understand things too well and perhaps counting too much on the fact that its implementation looked like it should be re-entrant. Of course I ended up with a little bit of a nightmare at hand.Maybe I should just be less stubborn and build yet another full-screen app instead.
In any case, here's some of the nightmarish stuff I have been able to cause thus far:
I understand that providing a minimal code snippet to reproduce these issues would be better, and perhaps I'll find the time to do so in the future. Meanwhile, is there any chance from the little info I could provide you'd have any hint as to how I should be doing instead and what I shouldn't be doing?
Thank you!
The text was updated successfully, but these errors were encountered: