[tmzullinger]

Since less tag v706¹, less exits with a failure status when no filename is provided. Pass the -? option and redirect both stdout and stderr to /dev/null.

Failure to detect less results in using more for users who don't have MANPAGER or PAGER set, which is unexpected.

¹ gwsw/less@7c85230 (Exit with a failure
status in more cases, 2026-06-14)

• Issue: pyrepl.pager: less detection is broken with less >= 706 #161039

[python-cla-bot]

All commit authors signed the Contributor License Agreement.

[bedevere-app]

Most changes to Python require a NEWS entry. Add one using the blurb_it web app or the blurb command-line tool.

If this change has little impact on Python users, wait for a maintainer to apply the skip news label instead.

[tmzullinger]

This could be handled in various ways. Using less -? seemed like a clear and easy way to go. It is supported by less for many, many years. (I didn't check the code history, but a system with less-436 shows the -? option there.)

[tmzullinger]

Apparently, I should not have bothered adding a NEWS snippet after the initial feedback from the bot. Or at the least, I don't know enough about the syntax to write one that passes the doc checks.

If someone can offer advice on fixing whatever is wrong with the snippet or whether it's even needed for the simple code fix, I'll happily change it either way.

[StanFromIreland]

I pushed a little change to test with less -V instead of less -?, there's no shell glob so it's a little safer IMO. I also slightly reworded the news entry to explain the user-facing issue.

[tmzullinger]

I pushed a little change to test with less -V instead of less -?, there's no shell glob so it's a little safer IMO. I also slightly reworded the news entry to explain the user-facing issue.

Thanks! I can update the commit message to match, if that level of consistency is important to anyone.

Indeed it's better to avoid -? in the case someone has a file with -<char> in the working directory. I should have thought about this due to the use of the shell via os.system. I used --help initially but wanted to shorten it so I didn't have to think more about how and where you might want the code wrapped. And less (being quite old) uses -h for other things.

(I did wonder why os.system was used instead of subprocess or something else. Is that intentional, perhaps to avoid more dependencies, or just quirk of how the pager code evolved?)

FWIW, I checked the less history and both -? and -V are both supported practically forever. -? was added in fe6f798 (Initial revision, 1994-10-15) (tag v243). -V was added a few weeks later in 0aae6ed (Add -V option., 1994-11-01) (tag v250). :)

[miss-islington-app]

Thanks @tmzullinger for the PR, and @StanFromIreland for merging it 🌮🎉.. I'm working now to backport this PR to: 3.14, 3.15.
🐍🍒⛏🤖

[StanFromIreland]

Merged, thanks!

Is that intentional, perhaps to avoid more dependencies, or just quirk of how the pager code evolved?

I presume the pager predates subprocess, and I don't think there's enough reason to switch currently.

Hmm, I'll try backporting again tomorrow, the bot seems to be down.

[miss-islington-app]

Thanks @tmzullinger for the PR, and @StanFromIreland for merging it 🌮🎉.. I'm working now to backport this PR to: 3.15.
🐍🍒⛏🤖

[miss-islington-app]

Thanks @tmzullinger for the PR, and @StanFromIreland for merging it 🌮🎉.. I'm working now to backport this PR to: 3.15.
🐍🍒⛏🤖

[bedevere-app]

GH-161148 is a backport of this pull request to the 3.15 branch.

[miss-islington-app]

Thanks @tmzullinger for the PR, and @StanFromIreland for merging it 🌮🎉.. I'm working now to backport this PR to: 3.14.
🐍🍒⛏🤖

[bedevere-app]

GH-161149 is a backport of this pull request to the 3.14 branch.