Re: win-test: unknown terminal "xterm-256color", was Re: [RFC PATCH 6/6] ls-tree: introduce '--pattern' option

From: Jeff King <peff@peff.net>
Date: Mar 28, 2023, 20:43 UTC
Message-ID: <20230328204318.GA1274197@coredump.intra.peff.net>
In-Reply-To: <20230328195942.GA1173031@coredump.intra.peff.net>

On Tue, Mar 28, 2023 at 03:59:42PM -0400, Jeff King wrote:

> At any rate, unconditionally setting TERM=dumb should cover all cases,
> I'd think. I'm happy to prepare a patch.

Hmph. So I started to do this, and notice there are a bunch of tput calls in the ci/ scripts themselves.

So if I do a dummy workflow like this:

-- >8 --
name: fake ci
on: [push, pull_request]
jobs:
  ubuntu:
    runs-on: ubuntu-latest
    steps:
      - name: check TERM
        run: |
          echo TERM=$TERM
          echo $(tput setaf 1)this is setaf 1$(tput sgr0)
  windows:
    runs-on: windows-latest
    steps:
      - name: check TERM
        shell: bash
        run: |
          echo TERM=$TERM
          echo $(tput setaf 1)this is setaf 1$(tput sgr0)
-- 8< --

then I note two interesting things:

  - the attempted colorization does nothing on ubuntu, because it's
    already TERM=dumb

  - the colorization on windows works! But I thought it was exactly
    these sorts of tputs that are causing the error messages.

So now I'm more confused than ever. I have a feeling we would be better off just using ansi codes instead of tput, as Johannes suggested earlier. But given the twists and turns here, and the fact that nobody really seems that bothered by the status quo, I'm inclined to just leave it alone for now.

-Peff