Top-level sage script: Unconditionally set SAGE_ROOT

[mkoeppe]

This will make it more robust when accidentally invoked from a sage shell inside another Sage installation.

Related to https://groups.google.com/g/sage-release/c/GOGWk66zaCQ/m/0o9KtVDIAAAJ

Depends on #33787

CC: @dimpase @orlitzky @jhpalmieri

Component: scripts

Author: Matthias Koeppe

Branch/Commit: 01caa38

Reviewer: John Palmieri

Issue created by migration from https://trac.sagemath.org/ticket/33786

[mkoeppe]

Dependencies: #33787

[mkoeppe]

Branch: u/mkoeppe/top_level_sage_script__unconditionally_set_sage_root

[mkoeppe]

Commit: 252372d

[mkoeppe]

Author: Matthias Koeppe

[mkoeppe]

New commits:

60bcbec	README.md, src/doc/en/installation/source.rst: Update multi-user install
2116f9f	README.md: Update final step 'symlink sage', add jupyter kernel step
5c68335	Merge #33787
252372d	SAGE_ROOT/sage: Unconditionally determine SAGE_ROOT from $0; no longer invite to edit this file

[jhpalmieri]

comment:5

It looks like this doesn't change the behavior of SAGE_ROOT/sage, just the documentation. Is that correct?

[mkoeppe]

comment:6

Looks like I forgot to push a commit

[sagetrac-git]

Changed commit from 252372d to d703cd1

[sagetrac-git]

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

d703cd1

SAGE_ROOT/sage: Actually unconditionally determine SAGE_ROOT from -zsh

[mkoeppe]

comment:8

Here it is

[sagetrac-git]

Changed commit from d703cd1 to 01caa38

[sagetrac-git]

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

01caa38

SAGE_ROOT/sage: Actually unconditionally determine SAGE_ROOT from $0

[jhpalmieri]

comment:10

How do we determine whether people are actually setting SAGE_ROOT outside of this file and/or modifying the commented-out line in this file? That is, how do we determine how many Sage installations this might break? Should we deprecate the setting of SAGE_ROOT first?

if [ -z "$SAGE_ROOT" ];  then
   ...
else
   (print deprecation warning)
fi

[mkoeppe]

comment:11

Replying to @jhpalmieri:

How do we determine whether people are actually setting SAGE_ROOT outside of this file and/or modifying the commented-out line in this file? That is, how do we determine how many Sage installations this might break?

I don't think this change can break anyone's installation if that was done using the old instructions ("system-wide install"):

• the first variant, using symlinks, still works (we just don't advertise it any more)

• in the second variant, using a copy of the script, there's no problem because users were instructed to make the change (setting SAGE_ROOT) in the copy, not in the original. Their old copy will continue to work. If they attempt to copy and make the change again, they will see that there is no longer a place that prompts them to make the change.

[jhpalmieri]

Reviewer: John Palmieri