__package_apt to accept .dpkg file as --source #66
Conversation
| name="${name}=${version_should}" | ||
| aptget="${aptget:?} -o APT::Get::allow-downgrades=true" | ||
|
|
||
| echo "${aptget} ${recommendsparam} install ${target_release} '${name}'" |
There was a problem hiding this comment.
Moving the install command up here breaks the auto-update mechanism below (line 121ff.)
There was a problem hiding this comment.
Fair point. I shall have a think about this aspect.
| if [ -f "$__object/parameter/source" ] | ||
| then | ||
| source_path="$(cat "$__object/parameter/source")" | ||
| tempfile="$($__remote_exec $__target_host "mktemp -t cidst.XXXXX.${source_path##*/}")" |
There was a problem hiding this comment.
I don't consider executing $__remote_* commands from a gencode-* script good style, as they will also be executed for dry runs.
This section should go to a code-local or, even better, use __file if possible.
Supporting --source -, like __file would also be nice, IMO.
also: cidst -> cdist
There was a problem hiding this comment.
Well, that's an interesting and important aspect. Thanks for pointing it out.
However, I am in a bit of a pickle here:
- Putting it into
gencode-localwont' really fly here: that will return a filename (and/or create a file) that's unique locally. - Putting it into
__filewould then mean, we need to pass the value from themanifestto thegencode-remotescript, which then means setting up temporary files and such, right?
Supporting --sourece - seems to be rather out of scope here: how many people will pipe a .dpkg as a string literal through here?
Typo is a nice catch, thanks.
sideeffect42
changed the title
2 participants
No description provided.