Proposal: Eliminate INST_CONDA=0 from install script
Hi all, If anyone followed the e-mail thread on yt-users between me and Carla Bernhardt, they'd see that we had a lot of difficultly dealing with the INST_CONDA=0 option in the install script. This option reverts the install script to "legacy" mode, installing yt and all of yt's dependencies from source rather than using miniconda and conda packages to bootstrap a python environment. The problem is that the packages that get installed must be manually and laboriously updated. In practice this means packages are basically never updated. This also means the script installs relatively old versions of libraries (currently, matplotlib 1.5.1, numpy 1.11) simply because no one has taken the time to go through the list of libraries installed and update them. This means anyone who uses this option isn't benefiting from upstream bugfixes. It also means we risk bitrot. In practice, I don't think I've recently heard of a situation where INST_CONDA=1 failed but INST_CONDA=0 did not. The other option of course is for someone to step up and take responsibility for updating the INST_CONDA=0 section of the install script. I'd be happy to help out with that if anyone wants to volunteer. However, without a maintainer, I think we are doing our users a disservice by leaving this option available. If we *do* eliminate this option, we immediately reap the following benefits: * The install script can be trimmed down by several hundred lines * We can simplify the install-script based install instructions in the docs * It becomes easier to test modifications to the install script because there are fewer permutations to test. Please let me know what you think. I'd especially like to hear from people who are fans of INST_CONDA=0 or who recently used it with no issues. -Nathan
Hi Nathan, Unexpectedly, I find myself in agreement with you. I think we should ditch INST_CONDA=0, now that yt is able to be installed in a multitude of ways, and now that the dependency stack is also much easier to install. This will also lighten the burden of maintenance that we put on ourselves when we add optional or explicit dependencies. -Matt On Tue, Jun 20, 2017 at 11:04 AM, Nathan Goldbaum <nathan12343@gmail.com> wrote:
this change sounds good to me. On Tue, Jun 20, 2017 at 12:09 PM, Matthew Turk <matthewturk@gmail.com> wrote:
-- Michael Zingale Associate Professor Dept. of Physics & Astronomy • Stony Brook University • Stony Brook, NY 11794-3800 *phone*: 631-632-8225 *e-mail*: Michael.Zingale@stonybrook.edu *web*: http://www.astro.sunysb.edu/mzingale github: http://github.com/zingale
I'm glad everyone is OK with this :) Here's a PR that implements this proposal: https://github.com/yt-project/yt/pull/1461 On Tue, Jun 20, 2017 at 12:25 PM, Alex Lindsay <alexlindsay239@gmail.com> wrote:
Hi Nathan, Unexpectedly, I find myself in agreement with you. I think we should ditch INST_CONDA=0, now that yt is able to be installed in a multitude of ways, and now that the dependency stack is also much easier to install. This will also lighten the burden of maintenance that we put on ourselves when we add optional or explicit dependencies. -Matt On Tue, Jun 20, 2017 at 11:04 AM, Nathan Goldbaum <nathan12343@gmail.com> wrote:
this change sounds good to me. On Tue, Jun 20, 2017 at 12:09 PM, Matthew Turk <matthewturk@gmail.com> wrote:
-- Michael Zingale Associate Professor Dept. of Physics & Astronomy • Stony Brook University • Stony Brook, NY 11794-3800 *phone*: 631-632-8225 *e-mail*: Michael.Zingale@stonybrook.edu *web*: http://www.astro.sunysb.edu/mzingale github: http://github.com/zingale
I'm glad everyone is OK with this :) Here's a PR that implements this proposal: https://github.com/yt-project/yt/pull/1461 On Tue, Jun 20, 2017 at 12:25 PM, Alex Lindsay <alexlindsay239@gmail.com> wrote:
participants (5)
-
Alex Lindsay -
John ZuHone -
Matthew Turk -
Michael Zingale -
Nathan Goldbaum