| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
| |
This cleans up the ds install command to use namespaced folders on
digital ocean to hold assets. This will make it easier to install things
in the future if we break it down by version, but really it just makes
it easier to not have to do weird splitting logic to choose the right
tool and download and install it. For some reason hooked was also not
formatted properly when rustfmt was run on OSX so this was fixed to
make the build not break.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Up to this point hooked had been only designed to work on Windows, not
because dev-suite didn't want to support it, but because doing so was an
immense amount of work with no clear design due to how Unix and Window
paths are not at all the same. While shebang notation works on them for
both the paths are different.
In order to get around this we wrap Ruby, Python, and Bash scripts on
Windows with a different script that invokes the 'Git for Windows'
sh.exe to run the actual interpreters on the script. These can work fine
then as long as one has installed Git for Windows on their machine, and
has a copy of py.exe or ruby.exe on their path to be invoked.
There is one caveat. We have to assume that a user has installed their
copy of Git for Windows in the default location. This means if they
haven't the scripts will fail to run. There's not much we can do about
this and it's just a necessary wart to provide cross platform
capabilities for a project.
All projects can be initialized now with one of the language choices and
then have the proper files linked on their OS as part of the
initialization. Those who need to just link them in an already existing
project can just run `hooked link` in order to set their computer up.
This again handles the differences between the platforms. This project
is also updated to the new format of hooked so that collaboration is now
not limited to just Unix based OSes.
|
| | |
|
| |
|
|
|
|
|
|
| |
This commit really ups the level and quality of the Rust code by setting
clippy to pedantic mode. It also fixes an issue where bash continued to
run scripts even if something failed with a non-zero exit status. We
also deny all warnings so as to actually fail the builds and the commit
hooks. This should make sure code quality stays at a high level.
|
| |
|
|
|
|
|
|
| |
We really should be logging what's going on as each of these tools run.
Before this change hooked just ran without any indication of what was
going on. This change adds logging with info level for the end user by
default, with debug and trace statements while developing the code being
an option as well.
|
| |
|
|
|
|
|
|
|
| |
Up to this point testing of our command line tools just hasn't been
happening. That's not great. While locally testing things by hand is
possible, overtime various workflows will be harder to test by hand. By
automating these tests we can avoid regressions that we wouldn't think
to catch. Future work will involve working on adding tests for tools as
they integrate together.
|
| |
|
|
|
|
|
|
|
| |
This adds a commit to handle git commit linting to enforce style by not
allowing less than 10 or more than 50 chars for titles and less than or
equal to 72 chars for the body. Chars are measured in number of
graphemes as 50 chars represented in the terminal is what we want to use
not 50 bytes. This will eventually be an installable hook for end users
if they want it.
|
|
|
This adds the hooked binary to the dev-suite repo as well as a stub for
a program to be used in this workflow! Hooked works by adding the hooks
into the repo and setting them to executable and linking them into the
hooks directory under .git. This means hooks get to travel with the
repo and are source controlled. All a dev needs to do is run the init
command and hooked will symlink them all for them. No need to remember
how ln works. It's all handled for you. Future work will iterate about
what hooks that dev-suite supplies as part of the script. This will
involve configuration files and per repo settings are something that
will need to be thought about.
Closes Issue #2
|
