UX optimization: reversed patch == hung fetch... about please HOT 5 OPEN

Yeah, that sounds like the right thing to do. Which build defs are you experiencing this with? There are a number of places where we patch things in Please.

from please.

I agree, we shouldn't be hanging on interactive input like this. Do you know what these flags do? I think we essentially want the default behavior (i.e. don't automatically reverse it for us) but when it detects this case, it should just error. From this description here, I assume -t would have applied this patch, a bit like if you passed -R:

  -t  --batch  Ask no questions; skip bad-Prereq patches; assume reversed.
  -f  --force  Like -t, but ignore bad-Prereq patches, and assume unreversed.

From a UX perspective, having the output updated to be "Patching" vs "Fetching" would've also been a sufficient clue.

This is fairly easy to do but has a performance hit as patches are often used to patch large repos. We'd have to move patching to it's own target so it can have a different building description. This means we have to copy the files into a temp directory, and adding more intermediate targets means we have to repeat this step for each target.

from please.

I'd use -f. If the patch fails to apply, it should be an error (a reversed patch shouldn't magically apply because this is a form of non-determinism). In my case, the -f would've caught this error and it would've been obvious there was a patch problem, not something blocked waiting for input on stdin.

from please.

go_module():

go_module(
    name = "tengo",
    install = [
        "",
        "parser",
        "stdlib",
        "stdlib/json",
        "token",
    ],
    module = "github.com/d5/tengo/v2",
    patch = "d5-tengo-recover.patch",
    version = "v2.16.1",
)

d5/tengo#439, for reference

from please.