-
Notifications
You must be signed in to change notification settings - Fork 31
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
add shell specific packages #158
Conversation
edb8cdf
to
0242bc8
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is the only way to get bash support by adding argcomplete? I would have thought that these would be separate notions (and that I could get prefix.bash files generated even if I don't want to/can't install argcomplete)
Either way, if it fixes the packaging, then +1.
I might be calling it incorrectly, but it doesn't seem like the powershell script works even on the windows build machines:
Are you testing the result of the windows packages somehow? |
edit: the machine I tested on (windshield) has colcon-powershell==0.0.0 while the machine the archive was built on (icecube) had 0.1.0, which might be related if there are runtime dependency minimums |
These two feature (bash and argcomplete) come in two separate packages. |
We are not using the PowerShell scripts yet. We are still relying on Batch. It has certainly worked in the past. Maybe it is a regression of colcon/colcon-powershell#2. |
Addressed by colcon/colcon-powershell#3. |
OK I have a better understanding now of the impact of the dependencies.
I didn't follow the reasoning for installing optional dependencies on the buildfarm machines so I assumed it was required. We were trying to keep the dependencies minimal, so if argcomplete isn't required, we should remove it from this PR then, right? Or is it added so that the archives have argcomplete support or something? (in the past that was done based on user-installed dependencies not build-machine deps)
Given that powershell is not supported yet (e.g. we don't have matching env hooks for powershell) I think it's misleading to provide the powershell scripts in the archives, and think the |
0242bc8
to
14f1ef4
Compare
I removed the
The fact that some packages don't have an environment hook atm is imo not a reason to remove it. We currently also don't recommend using the I think it should stay in order to actually be able to test and use it. And potentially add the missing environment hooks where necessary. |
I agree that this could be useful one day, but I don't agree with this statement:
Users will notice the file, just as I did, and try to use it, whether it is recommended or not. If it's shipped in our archives then yes, as a user, I would expect it to work. Since we ship archives with opensplice and I am of the understanding that opensplice support will not be available for powershell, having the system fail on a user in a strange way IMO is not an acceptable trade-off so that we can hypothetically test powershell at some time in the future. If there isn't full support for powershell with the packages that are included in the archive, then the archive should not include Easiest way to achieve that is to not install the dependency on the machines. If we think that's unavoidable then perhaps the packaging job should remove the file manually. I don't believe this applies to zsh for example, because it wraps the sh env hooks from my understanding. If this is not correct, then we should remove the zsh file too is support is not complete. |
We are talking about an imminent need to use PowerShell because of the limitations of the |
OK, sure, if we plan on adding full powershell support soon then we can leave it as is. If the next release rolls around and it's not complete, I suggest we remove the .ps1 file. |
Fixes #157. Connect to #157.