bootstrap, src/doc/Makefile: Remove use of the SAGE_ROOT/sage script

Question

bootstrap, src/doc/Makefile: Remove use of the SAGE_ROOT/sage script

mkoeppe opened this issue 3 years ago · 34 comments

We change the bootstrap scripts so that they invoke build/bin/sage-package directly instead of going through the SAGE_ROOT/sage script.

This is faster and also more robust because it separates build phases better by eliminating the (unnecessary) use of configure-generated files such as src/bin/sage-src-env-config.

We also move the generation of src/doc/en/reference/repl/options.txt (by calling sage -advanced) from bootstrap time to src/doc/Makefile. The information displayed there depends on what packages are installed, so running it at bootstrap time was not correct.

Depends on #33740

CC: @jhpalmieri @tobiasdiez @orlitzky @dimpase @kiwifb

Component: scripts

Author: Matthias Koeppe

Branch/Commit: 67c8238

Reviewer: John Palmieri, François Bissey

Issue created by migration from https://trac.sagemath.org/ticket/33852

Answer 1 · 2022-05-14T21:08:41.000Z

Branch: u/mkoeppe/bootstrap__remove_use_of_the_sage_root_sage_script

Answer 2 · 2022-05-14T21:13:56.000Z

Commit: ccbc21a

Answer 3 · 2022-05-14T21:13:56.000Z

Last 10 new commits:

`10338a7`	`Change ci as well`
`0cf7334`	`Add sage-setup as install-requires dep`
`6d2b149`	`Use relative paths to not install src package`
`1dba3ba`	`Maybe thats the correct shebang?`
`c1b2fbf`	`Revert install_requires change`
`ad56ba3`	`Go back to two pip installs for now`
`b42f5f5`	`Readd how to use specific python version`
`87f360c`	`Readd name arg`
`765dff7`	`Merge #33740`
`ccbc21a`	`bootstrap, bootstrap-conda, src/doc/bootstrap: Use build/bin/sage-package directly`

Answer 4 · 2022-05-14T21:23:50.000Z

Changed commit from ccbc21a to 4560895

Answer 5 · 2022-05-14T21:23:50.000Z

Branch pushed to git repo; I updated commit sha1. New commits:

`4560895`	`src/doc/Makefile: Remove explicit use of SAGE_ROOT and the top-level sage script`

Answer 6 · 2022-05-14T21:37:04.000Z

Changed commit from 4560895 to 7759f3c

Answer 7 · 2022-05-14T21:37:04.000Z

Branch pushed to git repo; I updated commit sha1. New commits:

`7759f3c`	`Generate en/reference/repl/options.txt at build time, not bootstrap time`

Answer 8 · 2022-05-14T21:39:35.000Z

Description changed:

--- 
+++ 
@@ -2,3 +2,5 @@
 
 This is faster and also more robust because it separates build phases better by eliminating the (unnecessary) use of `configure`-generated files such as `src/bin/sage-src-env-config`.
 
+We also move the generation of `src/doc/en/reference/repl/options.txt` (by calling `sage -advanced`) from bootstrap time to `src/doc/Makefile`. The information displayed there depends on what packages are installed, so running it at bootstrap time was not correct.
+

Answer 9 · 2022-05-15T00:46:20.000Z

Branch pushed to git repo; I updated commit sha1. New commits:

`67c8238`	`src/bin/sage: Print error to error output`

Answer 10 · 2022-05-15T00:46:20.000Z

Changed commit from 7759f3c to 67c8238

Answer 11 · 2022-05-15T02:39:07.000Z

Branch pushed to git repo; I updated commit sha1. New commits:

`c671cbb`	`m4/sage_spkg_collect.m4: Read files dependencies_{optional,order_only,check}`
`9f3a296`	`build/pkgs: Split dependencies_optional and dependencies_check out from dependencies`

Answer 12 · 2022-05-15T02:39:07.000Z

Changed commit from 67c8238 to 9f3a296

Answer 13 · 2022-05-15T03:06:01.000Z

Changed commit from 9f3a296 to 3faf0ab

Answer 14 · 2022-05-15T03:06:01.000Z

Branch pushed to git repo; I updated commit sha1. New commits:

`46144d1`	`build/bin/sage-spkg-info: Restore message '(none known)' for packages without system package info`
`3faf0ab`	`build/bin/sage-spkg-info: Format dependencies`

Answer 15 · 2022-05-15T03:23:58.000Z

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

`facb1fc`	`build/bin/sage-spkg-info: Format dependencies`

Answer 16 · 2022-05-15T03:23:58.000Z

Changed commit from 3faf0ab to facb1fc

Answer 17 · 2022-05-15T03:26:52.000Z

Branch pushed to git repo; I updated commit sha1. New commits:

`8ab0814`	`build/pkgs/memory_allocator/SPKG.rst: Fix markup`

Answer 18 · 2022-05-15T03:26:52.000Z

Changed commit from facb1fc to 8ab0814

Answer 19 · 2022-05-15T03:32:06.000Z

Changed commit from 8ab0814 to 2d12c0b

Answer 20 · 2022-05-15T03:32:06.000Z

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

`2d12c0b`	`build/pkgs/memory_allocator/SPKG.rst: Fix markup`

Answer 21 · 2022-05-15T04:11:54.000Z

Branch pushed to git repo; I updated commit sha1. New commits:

`0e5162a`	`build/pkgs/mathics/SPKG.rst: Fix title`
`51eb305`	`build/bin/sage-spkg-info: Fix rst label`

Answer 22 · 2022-05-15T04:11:54.000Z

Changed commit from 2d12c0b to 51eb305

Answer 23 · 2022-05-15T04:15:58.000Z

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

Answer 24 · 2022-05-15T04:15:58.000Z

Changed commit from 51eb305 to 67c8238

Answer 25 · 2022-05-26T20:11:28.000Z

comment:17

I found it a bit hard to figure out what came from #33740 and what is this ticket :(

Once I get past that issue, I agree with the changes. It is much cleaner and simpler. May not go far enough from a sage-on-gentoo point of view (I will still patch but we had that discussion sage --docbuild vs python -m sage_docbuild somewhere else and that step is not ready yet). Just one remark, all the targets are phony in src/doc/Makefile by necessity but we could have had en/reference/repl/options.txt as a real target instead of a phony one, not that it matter much in that sea of stuff that rebuild every time you call make.

Answer 26 · 2022-05-26T20:13:24.000Z

comment:18

Replying to @kiwifb:

we could have had en/reference/repl/options.txt as a real target instead of a phony one

Yes but what would be its dependencies?

Answer 27 · 2022-05-26T20:22:12.000Z

comment:19

Replying to @mkoeppe:

Replying to @kiwifb:

we could have had en/reference/repl/options.txt as a real target instead of a phony one

Yes but what would be its dependencies?

I am not sure I follow. It doesn't need to have dependencies or have I forgotten something fundamental about makefiles.

Answer 28 · 2022-05-26T20:28:19.000Z

comment:20

In that case it would never be updated, which can't be right

Answer 29 · 2022-05-26T20:58:12.000Z

comment:21

Replying to @mkoeppe:

In that case it would never be updated, which can't be right

Thank you for reminding me of some essentials :) I'd like to brush it off as it being too early in the morning to make a proper comment but I think that's just part of a pattern where I am too deep on the packager side, and thinking of workflow in which ultimately, makefile such as this one are "single use", and when something doesn't work you just restart from the top because you need the flow to be right. I am not spending enough time on the pure dev side, that's what

Answer 30 · 2022-05-27T18:42:01.000Z

comment:22

I think this looks okay, too. François, do you agree with a positive review?

Answer 31 · 2022-05-27T20:13:00.000Z

comment:23

Yes, let's get it in the next beta.

Answer 32 · 2022-05-28T15:07:58.000Z

Reviewer: John Palmieri, François Bissey

Answer 33 · 2022-05-28T17:24:35.000Z

comment:25

Thanks!

Answer 34 · 2022-06-06T21:24:19.000Z

Changed branch from u/mkoeppe/bootstrap__remove_use_of_the_sage_root_sage_script to 67c8238