Update TODO wrt code duplication and the hooks directory
This commit is contained in:
parent
e6ca958418
commit
075da536d3
43
TODO
43
TODO
@ -101,20 +101,23 @@ Bugs to fix and features to add for 5.0
|
||||
|
||||
* Clean up the hooks directory
|
||||
|
||||
Currently we have one directory for Ubuntu and Fedora named after
|
||||
the first supported release which is then duplicated for each
|
||||
subsequent release.
|
||||
We still have a link farm for hooks and a meta link farm for distro
|
||||
releases created on "make install". It probably would be better if
|
||||
we would just have one directory per distro (like with debian) but
|
||||
without the need to created symlinks on "make install"
|
||||
|
||||
Currently CentOS's 25-setup-kernel creates an fstab and
|
||||
90-make-fstab does again. It works, but that cries for debugging
|
||||
hell.
|
||||
|
||||
80-install-kernel is not yet merged into one hook script.
|
||||
|
||||
Common oneliners for code deduplication in the hooks/ directory:
|
||||
|
||||
$ find -L . -not -xtype l -not -type d -not -path '*/common/*' | sort -t / -k3
|
||||
$ fdupes -r1 . | sort -t / -k3
|
||||
$ find . -type f | sim_text -ipTt 50 | tac | column -t
|
||||
|
||||
I'm sure this won't scale forever. So to minimise code duplication
|
||||
I'd like to have one common directory per distribution (e.g. called
|
||||
ubuntu-common, fedora-common, debian-common, maybe even deb-common,
|
||||
rpm-common or yum-common for dpkg/apt-, rpm/yum-based
|
||||
distributions, etc.) with generic hooks valid for all or most of
|
||||
the releases of one distribution and then one hook directory per
|
||||
release (e.g. called ubuntu-10.04 or ubuntu-lucid or so) which has
|
||||
symbolic links to everything which can be used unchanged from the
|
||||
common directory and new files for everything which has to be
|
||||
different or only there.
|
||||
|
||||
* Create users, add ssh pubkeys to .ssh/authorized_keys
|
||||
|
||||
@ -161,19 +164,7 @@ Bugs to fix and features to add for 5.0
|
||||
* Code Deduplication / Refactor the code for less code duplication
|
||||
|
||||
bin/x* currently contain the same or similar code like e.g. in the
|
||||
function readConfigurationFile. This needs to be cleaned up
|
||||
|
||||
hooks/ currently contain the same or similar code. First identical
|
||||
stuff needs to be merged into one file plus symlinks to retain
|
||||
backwards compatibility while keeping the code maintainable. Next
|
||||
step would be to have just one file for similar stuff with a few
|
||||
if-then-elses in there.
|
||||
|
||||
Common oneliners for code deduplication in the hooks/ directory:
|
||||
|
||||
$ find -L . -not -xtype l -not -type d -not -path '*/common/*' | sort -t / -k3
|
||||
$ fdupes -r1 . | sort -t / -k3
|
||||
$ find . -type f | sim_text -ipTt 50 | tac | column -t
|
||||
function readConfigurationFile. This needs to be cleaned up.
|
||||
|
||||
* Use Perl::Critic
|
||||
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user