Re: charm-helpers fetch apt_install optimization
Jay, I think you're touching on some important topics. Tuning and optimizing charm installation without fetching tons of extra packages around an install is certainly something I'm very open to experimentation with. This particular merge makes me suggest we run this through the gauntlet of testing scenarios - as if we merge this as is it effects the default expected behavior of the apt_install, with 200 charms in the store - and im' going to venture to guess ~ 40% of those are python and charm-helpers based charms. This means a very expensive test would follow to ensure we don't break whats in the wild. If you were to refactor this to make that a parameter on apt_install, and leave the default behavior as is - I would be more confident in making the merge happen. All the best, Charles On Wed, Nov 19, 2014 at 6:27 PM, Jay Wren jay.w...@canonical.com wrote: Greetings, I wanted to bring attention to a pending MR https://code.launchpad.net/~evarlast/charm-helpers/use-no-install-recommends/+merge/242270 which I feel can help charm-helpers. By not installing recommended packages, charms will install packages faster, with less unneeded packages. This leads to faster deploys and faster experience for new users and an overall better feeling about juju. On a similar subject, I am asking for feedback from charm-helpers users regarding apt related activity. I feel it might make sense use apt_pkg for as many operations as possible. Would a merge request of this nature be considered or is fork and exec to the more familiar system commands preferred? Thanks, -- Jay -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju
Re: charm-helpers fetch apt_install optimization
I have to disagree slightly, currently (and thankfully?) the current way most charm-helpers aim to be included in charms is by embedding a working snapshot of that version of charm-helpers. So merging this wouldn't affect a huge number of charms, instead maybe the 5 or 6 at most that download from pypi and of those none may actual be broken. As we rethink the distribution model for charm-helpers (making it a debian package, having actual releases, etc) this would certainly become a larger concern. The changes look good at my cursory check, and I hope to review in more detail later this week. That said I just wanted to point out that I don't think a change, even as drastic as this would have a huge effect on the ecosystem. As charm authors sync charm-helpers they'll end up in our testing environment and should be caught then. Marco On Thu Nov 20 2014 at 12:33:55 PM Charles Butler charles.but...@canonical.com wrote: Jay, I think you're touching on some important topics. Tuning and optimizing charm installation without fetching tons of extra packages around an install is certainly something I'm very open to experimentation with. This particular merge makes me suggest we run this through the gauntlet of testing scenarios - as if we merge this as is it effects the default expected behavior of the apt_install, with 200 charms in the store - and im' going to venture to guess ~ 40% of those are python and charm-helpers based charms. This means a very expensive test would follow to ensure we don't break whats in the wild. If you were to refactor this to make that a parameter on apt_install, and leave the default behavior as is - I would be more confident in making the merge happen. All the best, Charles On Wed, Nov 19, 2014 at 6:27 PM, Jay Wren jay.w...@canonical.com wrote: Greetings, I wanted to bring attention to a pending MR https://code.launchpad.net/~evarlast/charm-helpers/use-no-install-recommends/+merge/242270 which I feel can help charm-helpers. By not installing recommended packages, charms will install packages faster, with less unneeded packages. This leads to faster deploys and faster experience for new users and an overall better feeling about juju. On a similar subject, I am asking for feedback from charm-helpers users regarding apt related activity. I feel it might make sense use apt_pkg for as many operations as possible. Would a merge request of this nature be considered or is fork and exec to the more familiar system commands preferred? Thanks, -- Jay -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/ mailman/listinfo/juju -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju
Re: charm-helpers fetch apt_install optimization
Reiterating my comment on the MP, I agree with Marco that, due in part to the fixed-point usage pattern of charm-helpers and in part to the specifics of the change, I don't see it as having much potential for breakage. That said, I'm certainly not against testing it before merging. On Thu, Nov 20, 2014 at 1:25 PM, Marco Ceppi ma...@ondina.co wrote: I have to disagree slightly, currently (and thankfully?) the current way most charm-helpers aim to be included in charms is by embedding a working snapshot of that version of charm-helpers. So merging this wouldn't affect a huge number of charms, instead maybe the 5 or 6 at most that download from pypi and of those none may actual be broken. As we rethink the distribution model for charm-helpers (making it a debian package, having actual releases, etc) this would certainly become a larger concern. The changes look good at my cursory check, and I hope to review in more detail later this week. That said I just wanted to point out that I don't think a change, even as drastic as this would have a huge effect on the ecosystem. As charm authors sync charm-helpers they'll end up in our testing environment and should be caught then. Marco On Thu Nov 20 2014 at 12:33:55 PM Charles Butler charles.but...@canonical.com wrote: Jay, I think you're touching on some important topics. Tuning and optimizing charm installation without fetching tons of extra packages around an install is certainly something I'm very open to experimentation with. This particular merge makes me suggest we run this through the gauntlet of testing scenarios - as if we merge this as is it effects the default expected behavior of the apt_install, with 200 charms in the store - and im' going to venture to guess ~ 40% of those are python and charm-helpers based charms. This means a very expensive test would follow to ensure we don't break whats in the wild. If you were to refactor this to make that a parameter on apt_install, and leave the default behavior as is - I would be more confident in making the merge happen. All the best, Charles On Wed, Nov 19, 2014 at 6:27 PM, Jay Wren jay.w...@canonical.com wrote: Greetings, I wanted to bring attention to a pending MR https://code.launchpad.net/~evarlast/charm-helpers/use-no-install-recommends/+merge/242270 which I feel can help charm-helpers. By not installing recommended packages, charms will install packages faster, with less unneeded packages. This leads to faster deploys and faster experience for new users and an overall better feeling about juju. On a similar subject, I am asking for feedback from charm-helpers users regarding apt related activity. I feel it might make sense use apt_pkg for as many operations as possible. Would a merge request of this nature be considered or is fork and exec to the more familiar system commands preferred? Thanks, -- Jay -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju
charm-helpers fetch apt_install optimization
Greetings, I wanted to bring attention to a pending MR https://code.launchpad.net/~evarlast/charm-helpers/use-no-install-recommends/+merge/242270 which I feel can help charm-helpers. By not installing recommended packages, charms will install packages faster, with less unneeded packages. This leads to faster deploys and faster experience for new users and an overall better feeling about juju. On a similar subject, I am asking for feedback from charm-helpers users regarding apt related activity. I feel it might make sense use apt_pkg for as many operations as possible. Would a merge request of this nature be considered or is fork and exec to the more familiar system commands preferred? Thanks, -- Jay -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju