* Replaced one of the = with ~ to allow any patch releases to be
compatible
* Removed pinning from requests, honcho & Click since we're using the
more basic APIs
* Made these changes so that it's easier to install bench globally :')
* chore: typo
* chore: update project description
* feat: command to install dev-requirements.txt
Often applications have development or test specific requirements which
are not required in production.
- Add new command `bench setup dev-requirements`
- installs all `dev-requirements.txt` in app's root folder.
* refactor: remove duplicate function
* refactor: use `log` instead of print
* refactor: merge dev-requirement command
* feat: install dev-dependencies in get-app if dev
When developer mode is enabled install all dev dependencies too while
doing `get-app`
* fix: warn about --dev not supporting node
* feat: set default version for easy install to v13
* fix: python version based version picking fixed and reverted changes to node version in installation script
* fix: revert test version
Co-authored-by: gavin <gavin18d@gmail.com>
These changes are valid only for Py3.6+ and will remove bench CLI
support for Python 3.5 and lesser.
* Converted to f-strings using flynt
* Manual conversion of failed cases
* Simplification of logic
If an unshallow clone has to be updated, the only way to update it
without a reset will require an unshallow. This is because there may be
uncommitted or committed changes in any of the apps, although unfavoured
should be handled safely. The only way to update such repos is via the
user configured settings..or a ff only merge.
Biggest bottleneck is in the validation stage which checks if the app is
installed on any existing sites on the bench. instead of triggering
multiple list-apps for each sites, do one for all.
Changed from execv to execve with empty env dict because bench commands on macOS 10.14.6 PY3.9.0a4 fails with "Error while finding module specification for 'frappe.utils.bench_helper' (ModuleNotFoundError: No module named 'frappe')"
This error has come up in the past too. I'm now suspecting it has
something to do with the current process passing on it's environment
variables to the process that's going to replace it, which causes these
issues.
Note that direct execution of the commands without the bench wrapper
worked fine although via bench failed with the specified error
