-
-
Notifications
You must be signed in to change notification settings - Fork 452
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Unify how SAGE_ROOT and SAGE_LOCAL are normalized regarding symbolic links #29446
Comments
comment:1
Another way to see this problem directly:
Rather than changing the doctest to use abspath, it seems that in both cases I am not sure where |
comment:2
Maybe it's just not a good doctest. Why do we care what the Python executable thinks SAGE_ROOT is? The top-level Note that this doctest is new as of 9.1.beta9. Maybe we should revert back to something closer to what was in beta8. |
comment:3
Replying to @jhpalmieri:
I assume this is a step towards making sagelib more usable from Python. Previously, SAGE_ROOT was just
Yes, that works as usual. |
comment:5
As of 9.2.beta3, I get a lot more test failures when building Sage in a location that is a symbolic link. The problem is that SAGE_ROOT changes between the symbolic link and the real path. For example:
|
comment:6
I think it needs to be investigated whether the |
comment:7
Just for the record, the new behaviour is a result of #25486. |
comment:8
That's right. |
comment:10
|
comment:12
Replying to @mkoeppe:
It seems that I think the problem is that Should this problem be handled in |
comment:13
Replying to @mwageringel:
From the comments at the top of
|
comment:14
Replying to @mwageringel:
Are you saying that the only setting in which you observe the reported warnings is when you manually call Then I'm not sure if anything needs to be fixed. |
comment:15
In
In
Of course, these comments are not precise. The In contrast, in
autoconf's |
comment:16
Also, in
|
comment:17
I see two possible fixes: a) Get rid of all the above b) |
comment:18
Replying to @mkoeppe:
Replying to @mkoeppe:
I prefer option b), since with a) it is confusing if the behavior depends on the (symbolic) path that was used to cd into SAGE_ROOT. If, for example, I open a new tab from SAGE_ROOT, the current directory is always the physical path, so it is easy to confuse the two. Option b) seems to work for me, but to test this on the affected machine I first need to install dependencies for bootstrap. |
Author: Markus Wageringel |
Branch: u/gh-mwageringel/29446 |
comment:19
Ok, this works for me now. With this branch, SAGE_ROOT will always be the absolute physical path, like it used to be until recently. In particular, the doctest from the description works now. SAGE_LOCAL was already the physical path, so nothing needs to be changed for that. New commits:
|
Commit: |
This comment has been minimized.
This comment has been minimized.
Reviewer: Matthias Koeppe |
comment:21
Thank you. |
Changed branch from u/gh-mwageringel/29446 to |
This ticket changes
SAGE_ROOT
, so that it is always the absolute physical path, i.e. symbolic links are resolved.As reported on sage-release, when building Sage 9.1.beta9 from a path that is a symbolic link, the following test in
src/sage/env.py
fails:The problem is that the left path in
out
is the symbolic link to Sage's root directory, whereasSAGE_ROOT
is the absolute path without symbolic link.CC: @mkoeppe @orlitzky
Component: build
Author: Markus Wageringel
Branch/Commit:
20f564e
Reviewer: Matthias Koeppe
Issue created by migration from https://trac.sagemath.org/ticket/29446
The text was updated successfully, but these errors were encountered: