Add support for heredocs in CMD dockerfile commands #2265
Conversation
|
Perhaps for this one we should also consider to waiting until we get more use-cases. I can see the benefit of being consistent with If this can be added in future without breaking backward-compatibility, I think it's ok to wait until we have practical use-cases. |
|
Closing for now to keep the heredoc scope from growing too fast. We can revisit if there are more use-cases. |
|
Thanks @jedevc ! |
jedevc
force-pushed
the
dockerfile-heredocs-cmd
branch
2 times, most recently
from
9a36ab6
to
84f25a5
Compare
This patch introduces support for using heredocs for the CMD instruction - they work identically to the heredocs for RUN, with a couple of differences for shebang scripts: - We create a new layer writing the script contents to /mnt/pipes/ Usually CMD does not create a new layer, however, the file is required to run the script as we would with RUN, since we want to avoid manual parsing of the shebang line. - We have to use /mnt/pipes instead of /dev/pipes, since /dev/pipes is mounted at runtime. Signed-off-by: Justin Chadwell <[email protected]>
jedevc
force-pushed
the
dockerfile-heredocs-cmd
branch
from
84f25a5
to
48cfb7e
Compare
|
I see @tonistiigi left some comments on the other PR; #3082 (comment) I'm personally also still on the fence if we should do this; it still feels like a corner-case, and can already be (largely) resolved by using Should we close this one again? |
As stated, this patch introduces allowing heredocs in the
CMDcommand exactly as in theRUNcommand 🎉I don't think this is particularly complex - although I do have one bit I'm not quite sure about. For the case of supporting shebangs, we can't have run-mounts for the
CMD, so I've gone ahead and introduced a new copying layer in these cases (into/mntthough, since/devisn't preserved when running). I'm happy to switch to one of the following though if that doesn't work:CMDCMDCMD