ROOT: new versions with associated dependency constraints by HadrienG2 · Pull Request #34185 · spack/spack

HadrienG2 · 2022-11-29T16:13:39Z

@wdconinc Here's my take on the dependency constraints, so that the info on your closed PR don't get lost :)

I couldn't spell out the macOS conflict because I don't have access to a Mac so I can't check how macOS fills uname information.

Feel free to test/approve this once you have the time.

wdconinc · 2022-11-29T16:33:24Z

As commented here (simultaneously to this PR), doesn't it make more sense to have this modeled as closed-ended dependencies rather than open-ended dependencies with conflicts? That would avoid 'automatically' allowing for example new major python versions when they come out even for older ROOT versions. It would 'pin' the dependencies to a known major version until a certain ROOT version that has consciously added support for it.

HadrienG2 · 2022-11-29T16:44:28Z

I'm not sure about full close-ended dependencies because it introduces significant extra maintenance effort where most dependency updates are actually harmless. But I pushed a version that avoids conflicts where possible in favor of version dependency constraints (as far as I know, it's not currently possible to avoid conflicts for OS and compiler constraints).

HadrienG2 · 2022-11-29T16:54:11Z

...and using https://stackoverflow.com/questions/1777344/how-to-detect-mac-os-version-using-python as a template, I added a conflict for macOS 13+ on older ROOT, though again I cannot test it due to lack of suitable hardware.

vvolkl · 2022-11-29T17:03:45Z

Spack already knows the macOS version, it's probably easier to just call spack.operating_systems.mac_os.macos_version than deal with platform

Also, I'm currently trying to get macOS builds to work for key4hep, I can do some testing on our build machines.

HadrienG2 · 2022-11-29T17:10:39Z

That looks interesting, but in typical Python API documentation fashion, the docs claim that this function returns a "version object" without clarifying how such an object is actually constructed or used...

HadrienG2 · 2022-11-29T17:14:29Z

...oh, well, let's just copy-paste an incantation from another package that uses that API and hope that it's correct 🤷

wdconinc · 2022-11-29T20:52:40Z

Tested that this builds correctly on my system for 6.26.10, with the following variants:

4fmb3qq [email protected]~aqua~arrow+davix~dcache~emacs+examples~fftw~fits~fortran+gdml+gminimal~graphviz+gsl+http~ipo~jemalloc+math~memstat+minuit~mlp~mysql+opengl~oracle~postgres~pythia6~pythia8+python~qt4~r+roofit+root7+rpath~shadow~spectrum~sqlite+ssl~table+tbb+threads~tmva+unuran+vc+vdt+veccore~vmc~webgui+x+xml+xrootd build_system=cmake build_type=RelWithDebInfo cxxstd=17 patches=22af347,89294c4

(89294c4 is a local patch; system is gcc@12; relevant dependency: [email protected])

HadrienG2 · 2022-11-30T06:22:18Z

Then let's undraft this.

tldahlgren · 2022-11-30T19:51:23Z

Looks like some pod-related issues with one of the pipeline stacks:

Waiting for pod pipeline/runner-k3q2udsw-project-2-concurrent-349nrm to be running, status is Pending
	ContainersNotInitialized: "containers with incomplete status: [init-permissions]"
	ContainersNotReady: "containers with unready status: [build helper]"
	ContainersNotReady: "containers with unready status: [build helper]"
Running on runner-k3q2udsw-project-2-concurrent-349nrm via x86-v2-pub-gitlab-runner-7684c89569-5zzvv...

@spackbot run pipeline

spackbot-app · 2022-11-30T19:51:27Z

I've started that pipeline for you!

var/spack/repos/builtin/packages/root/package.py

tldahlgren · 2022-11-30T20:38:38Z

Same stack failed again (see https://gitlab.spack.io/spack/spack/-/jobs/4803378).

var/spack/repos/builtin/packages/root/package.py

sethrj · 2022-12-01T16:54:22Z

var/spack/repos/builtin/packages/root/package.py


    # Python
    depends_on("[email protected]:", when="+python", type=("build", "run"))
+    depends_on("[email protected]:3.10", when="@:6.26.09 +python", type=("build", "run"))


Should this be a conflict indicating an issue with newer pythons on older ROOT? I'm not sure it makes a difference to the concretizer these days, but semantically it might be more appropriate.

Historically, one difference between this syntax and conflicts has been that in the conflict case, the spack concretizer would not try older dependency versions and just give up stating that there is a conflict. If you are telling me that this is not an issue anymore with newer spack versions, I can turn those back into conflicts.

The new concretizer is not "greedy" like the old one, so I think it should work. The real expert is @alalazo who could perhaps share some insight into best practices here?

In general it's better to add "positive" directives like done here. Conflicts add penalties to our minimization and they might trigger more easily weird situations where, to avoid the penalty, Spack turns off some other default.

sethrj

I successfully installed 6.26.10 on macOS 13.

alalazo · 2022-12-05T11:55:09Z

There's a conflict to be solved here

sethrj · 2022-12-05T13:43:26Z

The conflict was my bad (#34244) so I've fixed it.

sethrj · 2022-12-05T22:53:49Z

@tldahlgren would you mind checking the box for a second reviewer? Since I added a commit it won't accept my review to qualify for merge.

tldahlgren

LGTM

sethrj · 2022-12-06T15:00:26Z

@spackbot run ci

sethrj · 2022-12-06T15:13:13Z

@spackbot run pipeline

spackbot-app · 2022-12-06T15:13:18Z

I've started that pipeline for you!

@s

* Add new root versions and associated dependency constraints * Please style guide * Avoid conflicts where possible * Untested prototype of macOS version detection * Fixes for macOS version prototype * More logical ordering * More correctness and style fixes * Try to use spack's macos_version * Add some forgotten @s * Actually, Spack can't build Python 3.6 anymore, and thus no older PyROOT Co-authored-by: Seth R. Johnson <[email protected]>

spackbot-app bot added conflicts new-version update-package labels Nov 29, 2022

spackbot-app bot requested review from drbenmorgan, greenc-FNAL and vvolkl November 29, 2022 16:14

spackbot-app bot added the dependencies label Nov 29, 2022

HadrienG2 marked this pull request as ready for review November 30, 2022 06:22

HadrienG2 added 10 commits November 30, 2022 07:22

Add new root versions and associated dependency constraints

592ac25

Please style guide

335b93a

Avoid conflicts where possible

de7e93e

Untested prototype of macOS version detection

86b57fa

Fixes for macOS version prototype

d50b8e8

More logical ordering

eb5153b

More correctness and style fixes

6b042c2

Try to use spack's macos_version

710f9e7

Add some forgotten @s

7be650b

Actually, Spack can't build Python 3.6 anymore, and thus no older PyROOT

27823fe

tldahlgren assigned wdconinc and vvolkl Nov 30, 2022

tldahlgren reviewed Nov 30, 2022

View reviewed changes

var/spack/repos/builtin/packages/root/package.py Show resolved Hide resolved

drbenmorgan mentioned this pull request Dec 1, 2022

ROOT: add math/gsl conflict and change version-dependent features to conditional variants #34244

Merged

sethrj reviewed Dec 1, 2022

View reviewed changes

sethrj previously approved these changes Dec 2, 2022

View reviewed changes

Merge branch 'develop' into root-updates

37558c0

Merge remote-tracking branch 'upstream/develop' into root-updates

64e358f

sethrj dismissed their stale review via 64e358f December 5, 2022 13:41

sethrj approved these changes Dec 5, 2022

View reviewed changes

sethrj requested a review from tldahlgren December 5, 2022 20:39

tldahlgren assigned sethrj Dec 5, 2022

tldahlgren approved these changes Dec 5, 2022

View reviewed changes

sethrj enabled auto-merge (squash) December 5, 2022 23:10

sethrj changed the title ~~New root versions with associated dependency constraints~~ ROOT: new versions with associated dependency constraints Dec 5, 2022

sethrj merged commit 559c3de into spack:develop Dec 6, 2022

Conversation

HadrienG2 commented Nov 29, 2022

Uh oh!

wdconinc commented Nov 29, 2022

Uh oh!

HadrienG2 commented Nov 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

HadrienG2 commented Nov 29, 2022

Uh oh!

vvolkl commented Nov 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

HadrienG2 commented Nov 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

HadrienG2 commented Nov 29, 2022

Uh oh!

wdconinc commented Nov 29, 2022

Uh oh!

HadrienG2 commented Nov 30, 2022

Uh oh!

tldahlgren commented Nov 30, 2022

Uh oh!

spackbot-app bot commented Nov 30, 2022

Uh oh!

Uh oh!

tldahlgren commented Nov 30, 2022

Uh oh!

Uh oh!

sethrj Dec 1, 2022

Choose a reason for hiding this comment

Uh oh!

HadrienG2 Dec 1, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sethrj Dec 1, 2022

Choose a reason for hiding this comment

Uh oh!

alalazo Dec 2, 2022

Choose a reason for hiding this comment

Uh oh!

sethrj left a comment

Choose a reason for hiding this comment

Uh oh!

alalazo commented Dec 5, 2022

Uh oh!

sethrj commented Dec 5, 2022

Uh oh!

sethrj commented Dec 5, 2022

Uh oh!

tldahlgren left a comment

Choose a reason for hiding this comment

Uh oh!

sethrj commented Dec 6, 2022

Uh oh!

sethrj commented Dec 6, 2022

Uh oh!

spackbot-app bot commented Dec 6, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

HadrienG2 commented Nov 29, 2022 •

edited

Loading

vvolkl commented Nov 29, 2022 •

edited

Loading

HadrienG2 commented Nov 29, 2022 •

edited

Loading

HadrienG2 Dec 1, 2022 •

edited

Loading