remote + build + RUN: emits spurious CR (^M)

[edsantiago]

Another variant of #7942 : RUN commands within a podman build are emitting an unwanted ^M.

In window 1:

$ bin/podman system service --timeout=0

In window 2:

$ printf "FROM alpine\nRUN echo hi\n" | bin/podman-remote build -t foo - | cat -vET

Output is below, contrasting podman (left) to podman-remote (right):

STEP 1: FROM alpine$                | STEP 1: FROM alpine$
STEP 2: RUN echo hi$                | STEP 2: RUN echo hi$
hi$                                 | hi^M$   <---- unwanted ^M
STEP 3: COMMIT foo$                 | STEP 3: COMMIT foo$

master @ 4eb9c28; also present in podman-2.1.1-7.fc32

If/when this is fixed, please update test:

podman/test/system/070-build.bats

Lines 358 to 367 in 9fa09a8

	# FIXME: podman-remote output differs from local: #8342 (spurious ^M)
	# FIXME: podman-remote output differs from local: #8343 (extra SHA output)
	remote_extra=""
	if is_remote; then remote_extra=".*";fi
	expect="${random1}
	.*
	STEP 1: FROM $IMAGE
	STEP 2: RUN echo x${random2}y
	x${random2}y${remote_extra}
	STEP 3: COMMIT build_test${remote_extra}

[rhatdan]

@nalind Any ideas what is causing this?

[nalind]

Is there a pseudo-terminal involved? If so, does it have the "ONLCR" flag set on its output? Running "stty -a" in that context would tell us, either with "onlcr" (set) or "-onlcr" (not set).

[edsantiago]

The terminal from which I'm running has onlcr; if I try stty -onlcr life is too horrible to contemplate.

The ^M happens even if I s/|/|&/, which presumably removes ptys completely out of the equation:

$ printf "FROM alpine\nRUN echo hi\n" | bin/podman-remote build -t foo - |&  cat -vET
...
hi^M$

[nalind]

Is the command indicated in the RUN instruction being executed with a pseudo-terminal, and if so, is "onlcr" set in it?

Experimentally, it looks like docker build always disables use of pseudo-terminals for RUN, and I'm thinking we should switch to doing the same. That would also make containers/buildah#2577 viable.

[edsantiago]

Duh, sorry, I misunderstood.

$ printf "FROM alpine\nRUN stty -a\n" | bin/podman-remote build -t foo - |& cat -vET
STEP 1: FROM alpine$
STEP 2: RUN stty -a$
speed 38400 baud;stty: standard input^M$
 line = 0;^M$
intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol = <undef>;^M$
eol2 = <undef>; swtch = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R;^M$
werase = ^W; lnext = ^V; flush = ^O; min = 1; time = 0;^M$
-parenb -parodd -cmspar cs8 -hupcl -cstopb cread -clocal -crtscts^M$
-ignbrk -brkint -ignpar -parmrk -inpck -istrip -inlcr -igncr icrnl ixon -ixoff^M$
-iuclc -ixany -imaxbel -iutf8^M$
opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0^M$
isig icanon iexten echo echoe echok -echonl -noflsh -xcase -tostop -echoprt^M$
echoctl echoke -flusho -extproc^M$
STEP 3: COMMIT foo$
...
$ printf "FROM alpine\nRUN stty -onlcr && echo hi\n" | bin/podman-remote build -t foo - |& cat -vET 
STEP 1: FROM alpine$
STEP 2: RUN stty -onlcr && echo hi$
hi$     <--- yippee!
STEP 3: COMMIT foo$
...

podman-non-remote for comparison:

$ printf "FROM alpine\nRUN stty -a\n" | bin/podman build -t foo - |& cat -vET
STEP 1: FROM alpine$
STEP 2: RUN stty -a$
stty: standard input: Not a tty$
Error: error building at STEP "RUN stty -a": error while running runtime: exit status 1$

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[rhatdan]

This is fixed in the main branch.