26 Aug, 2012
2 commits
-
The BSD version of `sed` apparently does not have the `-r` option. So,
eliminated the use of `sed` in this place altogather. -
The `-r` is not present on BSD version of sed (which is used on OSX).
12 Aug, 2012
1 commit
-
Use this command to revert an `antigen-update`.
11 Aug, 2012
7 commits
-
The memoized results of these functions cannot be saved in a global hash map,
becuase these functions are always run in a subshell. Changes to variables in
the subshell do not reflect in the main shell instance. We'd have to resort to
use the `$REPLY` ways returning the result from these functions, if they were to
be memoized. But that way of doing things sucks. Totally unreadable. So, just
not memoizing. -
The positional arguments can also be specified as keyword arguments.
10 Aug, 2012
10 commits
-
Apparently lines ending in an operator don't need a `\` to continue to next
line. Realised this thanks to @Tarrasch's pull request. -
Conflicts:
antigen.zsh -
There is a fourth boolean field in the record, which indicates wheather the
plugin has a local clone made. -
Renamed function to a temporary looking name and to more reflect its purpose.
Also, undefining the function once its purpose has been done.
31 Jul, 2012
1 commit
-
Local bundles, if are recorded with a `--branch` option, should *always* be
cloned.
27 Jul, 2012
1 commit
-
We force the `--no-local-clone` argument to `-bundle` command, when the given
repo url is an absolute local path and it is *not* a git repo. It doesn't make
sense to clone something which is not a repo!
26 Jul, 2012
3 commits
-
Using a new `make_local_clone` variable which is an internal field that better
represents whether we need a local clone. -
Keyword arguments of the syntax `--arg` are supported, not just `--arg=val`.
25 Jul, 2012
1 commit
-
The `-`'s were being replaced to `_`'s even in the value of the given argument,
whereas the intention was to just replace them in the name of the argument. This
is now fixed.
24 Jul, 2012
5 commits
-
The new `--no-local-clone` argument to -bundle command makes it so that if the
bundle specified is a local path, then no new clone is made. The bundle is
loaded directly from the given path. -
The position_args variable is now an array, instead of a string with argument
names split by a space. This feels much cleaner. -
Currently, a keyword style argument given to -bundle command requires a value,
like `--name=val`. With this, just a `--name` would imply `--name=true`
20 Jul, 2012
3 commits
-
Running an `antigen-update` will now print a --oneline git log of the changes
that are pulled. Helpful if interested in what has just been updated and a
glimpse on what can be expected. -
This will be used to make the -ensure-repo command output more detailed
information on the changes, if any, made to the clone.
18 Jul, 2012
4 commits
-
The directory names were being converted to urls and then compared to the
contents of the record. The problem with this is there could be directories with
extremely wierd names in there, not created by antigen. This could throw the
dir->url conversion off.So, with this, the urls from the record are converted to directory names and are
compared with the contents of repos/. Any directory unaccounted for will be
converted to a repo url, solely for the purpose of displaying to the user. After
confirmation, these directories are deleted, whether they represent a git clone
or not. -
The force argument makes the `-cleanup` command behave in a non-interactive
fashion, by deleting the unused clones without asking for a confirmation. -
Keeping the branch and the repo url as separate fields for each bundle entry,
seems to be only causing trouble. Now, the repo url and the branch are joined
togather with a `|` character and split only when needed.
17 Jul, 2012
2 commits
-
The contents of unused_clones is not a url anymore. Its more of a `url|branch`
like string.