ruby: export generic builder #55636

siers · 2019-02-12T12:11:52Z

Allow building any ruby, like ruby-build does.

Motivation for this change

I think it would it make sense to allow building (m)any ruby, like ruby-build does.
What do you think about this change, @manveru, @vrthra? (did I ping the wrong people? 😅)

Things done

alyssais · 2019-02-12T15:28:58Z

This is wonderful. I’d been wanting to get around to this. I’ll have some more substantial feedback later, but I definitely want to have this in Nixpkgs

manveru · 2019-02-13T07:45:10Z

I like this very much :)

siers · 2019-02-13T07:47:29Z

I thought thought that maybe the version should be called something more specific, if maybe that is to be changed, so backwards compatibility would be easier.

zimbatm · 2019-02-13T10:27:43Z

pkgs/development/interpreters/ruby/default.nix

@@ -198,6 +198,10 @@ let
    ) args; in self;

 in {
+  ruby_build_generic = { version, ... } @ args: generic (args // {


Suggested change

ruby_build_generic = { version, ... } @ args: generic (args // {

mkRuby = { version, ... } @ args: generic (args // {

mk or build are the name prefixes that are generally used by that produce derivations.

It would be nice is mkRuby could also be used subsequently below to make it clear how to use it.

zimbatm · 2019-02-13T10:28:01Z

Great idea!

Allow building any ruby, like ruby-build does.

siers · 2019-02-17T23:39:42Z

I renamed it to mkRuby and rebased the branch (incorrectly, initially).

alyssais · 2019-02-18T01:35:10Z

pkgs/development/interpreters/ruby/default.nix

@@ -97,7 +97,7 @@ let
          (import ./patchsets.nix {
            inherit patchSet useRailsExpress ops;
            patchLevel = ver.patchLevel;
-          })."${ver.majMinTiny}";
+          })."${ver.majMinTiny}" or [];


I think that, if we’re doing this, the existing rubies should be built using mkRuby, and their patches should be passed into it. (I’d be fine with getting rid of patchsets.nix and just inlining the patches here — there are a lot fewer patches than there used to be.)

Should I not use the patches if useRailsExpress is false within the mkDerivation?

siers · 2019-02-24T23:33:48Z

Patches inlined, mkRuby — used! I added versionNrs and versionStr, because it felt right.

siers · 2019-02-27T13:12:01Z

The builds should cache though, right? (By that I mean it should be possible to make them cache, since nothing really changed, since the version numbers get converted to the same end value.)

Also, ping!

manveru · 2019-02-27T13:43:51Z

pkgs/development/interpreters/ruby/default.nix

    version = rubyVersion "2" "6" "1" "";
    sha256 = {
      src = "1f0w37jz2ryvlx260rw3s3wl0wg7dkzphb54lpvrqg90pfvly0hp";
      git = "07gp7df1izw9rdbp9ciw4q5kq8icx3zd5w1xrhwsw0dfbsmmnsrj";
    };
+    # no Rails Express patchset yet (2019-01-30)


The patchset is now available at https://github.com/rvm/rvm/blob/master/patchsets/ruby/2.6.1/railsexpress

I just copypasted, but I can include it — sure. :)

Mic92 · 2019-02-27T14:37:40Z

pkgs/development/interpreters/ruby/default.nix

-  ruby_2_3 = generic {
-    version = rubyVersion "2" "3" "8" "";
+in rec {
+  mkRuby = { version ? null, versionNrs ? null, versionStr ? null, ... } @ args:


Could also add some documentation to doc/languages-frameworks/ruby.xml about this function?

siers · 2019-03-13T23:05:30Z

Inlining the patches, which happens from 7297606 to b7b4c44, makes the cache fail. Is that supposed to be that way? I don't entirely know why or should it be that way.
Here's the nix-diff of ruby_2_3 instantiations:

+ /nix/store/8738cn2080gjc66k3clg7ddjafgfiqa7-ruby-2.3.8.drv:{out}
• The input named `ruby-2.3.8` differs
  - /nix/store/n6sbal9nmmnb6f9qg53n7lnskadv2p1l-ruby-2.3.8.drv:{out}
  + /nix/store/w22ldvcmfzkl0ms9r346ls95rk5l4w3b-ruby-2.3.8.drv:{out}
  • The set of input names do not match:
      - source
  • The environments do not match:
      patches=''
          ←/nix/store/swd1jahgs16x7vc13pqyha8h3v4ra5fb-source/patches/ruby/2.3/head/railsexpress/01-skip-broken-tests.patch /nix/store/swd1jahgs16x7vc13pqyha8h3v4ra5fb-source/patches/ruby/2.3/head/railsexpress/02-improve-gc-stats.patch /nix/store/swd1jahgs16x7vc13pqyha8h3v4ra5fb-source/patches/ruby/2.3/head/railsexpress/03-display-more-detailed-stack-trace.patch←
      ''

Please advise, I've no idea. :)

alyssais · 2019-03-20T22:57:49Z

Inlining the patches, which happens from 7297606 to b7b4c44, makes the cache fail. Is that supposed to be that way? I don't entirely know why or should it be that way.

Possibly related; not sure -- why change from using rubyVersion to versionNrs and versionStr?

siers · 2019-03-20T23:36:58Z

It translates down to rubyVersion, but I myself as a user would like to use the same format as in `.ruby-version` — that's why. :)

…

On Thu., 21 Mar. 2019, 6:57 am Alyssa Ross, ***@***.***> wrote: > Inlining the patches, which happens from 7297606 to b7b4c44, makes the cache fail. Is that supposed to be that way? I don't entirely know why or should it be that way. Possibly related; not sure -- why change from using rubyVersion to versionNrs and versionStr? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#55636 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAOWi9OtyP-7jOT-AbZskQ8Mrirvmkhtks5vYrzxgaJpZM4a2Qj7> .

alyssais · 2019-03-21T00:17:32Z

It translates down to rubyVersion, but I myself as a user would like to use the same format as in `.ruby-version` — that's why. :)

Hmm. I'm not sure it's worth it. It makes the code really quite difficult to follow. Maybe if you could pull it out into some other function or something we could have the best of both worlds. I do appreciate the idea of having it be the same as .ruby-version.

siers · 2019-03-21T02:23:52Z

I could just move the versionStr -> rubyVersion function from mkRuby into a function with its own name so you could 'get' it without having to read the implementation. How 'bout that? :P

Do those rebuilds qualify this as non-mergable for now?

manveru · 2019-03-21T09:11:53Z

I've used rubyVersion = lib.replaceChars ["." "\n"] ["_" ""] ( lib.readFile .ruby-version); for that purpose so far.

alyssais · 2019-03-21T09:26:14Z

I could just move the `versionStr -> rubyVersion` function from `mkRuby` into a function with its own name so you could 'get' it without having to read the implementation. How 'bout that? :P

Sounds good.

Do those rebuilds qualify this as non-mergable for now?

Not necessarily. I'm hoping whatever's causing them will just disappear as we continue to work on the code, but even if they stick around we can still merge this.

siers · 2019-03-22T02:18:19Z

@manveru That looks like a list. The object that ought to be created is that weird set that pkgs/development/interpreters/ruby/ruby-version.nix creates.

maksar · 2020-10-15T10:20:17Z

I am not quite familiar with PR processes in NixPkgs. @siers Can you please explain why this was closed?

siers · 2020-10-15T10:31:23Z

because I don't have lots of dedicated computer volunteer time and so I couldn't support it

maksar · 2020-10-15T10:48:49Z

because I don't have lots of dedicated computer volunteer time and so I couldn't support it

Got it ;) Maybe I'll continue a bit later. Will use your fork, thanks.

siers requested review from alyssais and zimbatm as code owners February 12, 2019 12:11

GrahamcOfBorg added 6.topic: ruby 10.rebuild-darwin: 0 10.rebuild-linux: 0 labels Feb 12, 2019

zimbatm reviewed Feb 13, 2019

View reviewed changes

siers force-pushed the ruby-export-generic-builder branch 2 times, most recently from 5534e14 to 11cedf1 Compare February 17, 2019 21:50

ruby: export generic builder

b7b4c44

Allow building any ruby, like ruby-build does.

siers force-pushed the ruby-export-generic-builder branch from 11cedf1 to b7b4c44 Compare February 17, 2019 23:39

alyssais requested changes Feb 18, 2019

View reviewed changes

siers added 2 commits February 25, 2019 01:30

ruby: inline ruby patches file list, rm patchsets.nix

631ec4c

ruby: use mkRuby to define other rubies, add versionNrs/versionStr

7297606

GrahamcOfBorg requested a review from vrthra February 25, 2019 00:38

GrahamcOfBorg added 10.rebuild-darwin: 101-500 10.rebuild-linux: 501+ 10.rebuild-linux: 501-1000 and removed 10.rebuild-darwin: 0 10.rebuild-linux: 0 labels Feb 25, 2019

manveru reviewed Feb 27, 2019

View reviewed changes

Mic92 reviewed Feb 27, 2019

View reviewed changes

FRidh added this to Not ready in Staging Oct 24, 2019

FRidh moved this from Not ready to New in Staging Oct 24, 2019

FRidh moved this from New to Not ready in Staging Oct 24, 2019

siers closed this Jan 26, 2020

Staging automation moved this from WIP to Done Jan 26, 2020

Artturin mentioned this pull request Jun 26, 2022

ruby: Expose generic builder #173390

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ruby: export generic builder #55636

ruby: export generic builder #55636

siers commented Feb 12, 2019 •

edited

alyssais commented Feb 12, 2019

manveru commented Feb 13, 2019

siers commented Feb 13, 2019

zimbatm Feb 13, 2019

zimbatm commented Feb 13, 2019

siers commented Feb 17, 2019 •

edited

alyssais Feb 18, 2019

siers Feb 24, 2019

siers commented Feb 24, 2019

siers commented Feb 27, 2019

manveru Feb 27, 2019

siers Feb 27, 2019

Mic92 Feb 27, 2019

siers commented Mar 13, 2019

alyssais commented Mar 20, 2019 via email

siers commented Mar 20, 2019 via email

alyssais commented Mar 21, 2019 via email

siers commented Mar 21, 2019

manveru commented Mar 21, 2019 •

edited

alyssais commented Mar 21, 2019 via email

siers commented Mar 22, 2019

maksar commented Oct 15, 2020

siers commented Oct 15, 2020

maksar commented Oct 15, 2020

	ruby_build_generic = { version, ... } @ args: generic (args // {
	mkRuby = { version, ... } @ args: generic (args // {

ruby: export generic builder #55636

ruby: export generic builder #55636

Conversation

siers commented Feb 12, 2019 • edited

Motivation for this change

Things done

alyssais commented Feb 12, 2019

manveru commented Feb 13, 2019

siers commented Feb 13, 2019

zimbatm Feb 13, 2019

Choose a reason for hiding this comment

zimbatm commented Feb 13, 2019

siers commented Feb 17, 2019 • edited

alyssais Feb 18, 2019

Choose a reason for hiding this comment

siers Feb 24, 2019

Choose a reason for hiding this comment

siers commented Feb 24, 2019

siers commented Feb 27, 2019

manveru Feb 27, 2019

Choose a reason for hiding this comment

siers Feb 27, 2019

Choose a reason for hiding this comment

Mic92 Feb 27, 2019

Choose a reason for hiding this comment

siers commented Mar 13, 2019

alyssais commented Mar 20, 2019 via email

siers commented Mar 20, 2019 via email

alyssais commented Mar 21, 2019 via email

siers commented Mar 21, 2019

manveru commented Mar 21, 2019 • edited

alyssais commented Mar 21, 2019 via email

siers commented Mar 22, 2019

maksar commented Oct 15, 2020

siers commented Oct 15, 2020

maksar commented Oct 15, 2020

siers commented Feb 12, 2019 •

edited

siers commented Feb 17, 2019 •

edited

manveru commented Mar 21, 2019 •

edited