markupsafe 2.1.0 build 0 breaks jinja2<3

[saraedum]

Solution to issue cannot be found in the documentation.

• I checked the documentation.

Issue

markupsafe 2.1.0 bulid 1 added a run_constraint for jinja2 requiring it to be >=3.

However, build 0 does not have that constraint, so one can sometimes get broken environments, as reported by @davidaulicino for a sagelib=9.2 environment (reported outside of github.)

A short reproducer is the following:

mamba create -n deleteme jinja2=2.10.3 markupsafe=2.1
conda activate deleteme
python -c 'import jinja2'
ImportError: cannot import name 'soft_unicode' from 'markupsafe'

Should the build 0 be marked as broken? (I don't think it's worth to update the metadata here since the builds are otherwise identical.)

Installed packages

_libgcc_mutex             0.1                 conda_forge    conda-forge
_openmp_mutex             4.5                       2_gnu    conda-forge
bzip2                     1.0.8                h7f98852_4    conda-forge
ca-certificates           2022.5.18.1          ha878542_0    conda-forge
jinja2                    2.10.3                     py_0
ld_impl_linux-64          2.36.1               hea4e1c9_2    conda-forge
libffi                    3.4.2                h7f98852_5    conda-forge
libgcc-ng                 12.1.0              h8d9b700_16    conda-forge
libgomp                   12.1.0              h8d9b700_16    conda-forge
libnsl                    2.0.0                h7f98852_0    conda-forge
libuuid                   2.32.1            h7f98852_1000    conda-forge
libzlib                   1.2.12               h166bdaf_0    conda-forge
markupsafe                2.1.0           py310h5764c6d_0    conda-forge
ncurses                   6.3                  h27087fc_1    conda-forge
openssl                   3.0.3                h166bdaf_0    conda-forge
pip                       22.1.2             pyhd8ed1ab_0    conda-forge
python                    3.10.4          h2660328_0_cpython    conda-forge
python_abi                3.10                    2_cp310    conda-forge
readline                  8.1                  h46c0cb4_0    conda-forge
setuptools                62.3.2          py310hff52083_0    conda-forge
sqlite                    3.38.5               h4ff8645_0    conda-forge
tk                        8.6.12               h27826a3_0    conda-forge
tzdata                    2022a                h191b570_0    conda-forge
wheel                     0.37.1             pyhd8ed1ab_0    conda-forge
xz                        5.2.5                h516909a_1    conda-forge
zlib                      1.2.12               h166bdaf_0    conda-forge

Environment info

active environment : deleteme
    active env location : /tmp/ruth/micromamba/envs/deleteme
            shell level : 1
       user config file : /home/ruth/.condarc
 populated config files : /home/ruth/.condarc
          conda version : 4.12.0
    conda-build version : not installed
         python version : 3.10.4.final.0
       virtual packages : __linux=4.19.0=0
                          __glibc=2.28=0
                          __unix=0=0
                          __archspec=1=x86_64
       base environment : /tmp/ruth/micromamba  (writable)
      conda av data dir : /tmp/ruth/micromamba/etc/conda
  conda av metadata url : None
           channel URLs : https://conda.anaconda.org/conda-forge/linux-64
                          https://conda.anaconda.org/conda-forge/noarch
                          https://repo.anaconda.com/pkgs/main/linux-64
                          https://repo.anaconda.com/pkgs/main/noarch
                          https://repo.anaconda.com/pkgs/r/linux-64
                          https://repo.anaconda.com/pkgs/r/noarch
          package cache : /tmp/ruth/micromamba/pkgs
                          /home/ruth/.conda/pkgs
       envs directories : /tmp/ruth/micromamba/envs
                          /home/ruth/.conda/envs
               platform : linux-64
             user-agent : conda/4.12.0 requests/2.27.1 CPython/3.10.4 Linux/4.19.0-20-amd64 debian/10 glibc/2.28
                UID:GID : 5012:500
             netrc file : None
           offline mode : False

[saraedum]

(I can open the PR to mark this as broken, if a maintainer agrees.)

[xylar]

I'm not a maintainer but I agree, this would be really helpful!

[xylar]

@saraedum, hmm, I don't think the problem is here actually. It's that jinja2 < 3 shouldn't be using markupsafe >=2.1.0. So the fix would be to try to patch the problematic jinja2 < 3. I kind of thought this had been done, actually.

[xylar]

By me, actually:
conda-forge/conda-forge-repodata-patches-feedstock#238
I don't know why some cases are sneaking through. But @saraedum if you can figure out why, I can update the patch.

[xylar]

@saraedum, it seems like your jinja2 isn't actually coming from conda-forge. No channel is listed so it could be defaults or pypi or your own installation or ???

[xylar]

@saraedum, I see when I do:

mamba search --info jinja2=2.10.3

that on the main channel there has been no patching to make sure that markupsafe >=0.23,<2 as there has been on conda-forge:

jinja2 2.10.3 py_0
------------------
file name   : jinja2-2.10.3-py_0.conda
name        : jinja2
version     : 2.10.3
build       : py_0
build number: 0
size        : 101 KB
license     : 3-Clause BSD
subdir      : noarch
url         : https://repo.anaconda.com/pkgs/main/noarch/jinja2-2.10.3-py_0.conda
md5         : d1ee8fe7106f93ac20f6714b7f3270bc
timestamp   : 2019-10-07 15:40:05 UTC
dependencies: 
  - markupsafe >=0.23
  - python
  - setuptools

Looks like an issue that will need to be addressed with the Anaconda folks rather than here.

I did notice that I missed patching one file in conda-forge/conda-forge-repodata-patches-feedstock#238 so I'm going to put in a fix.

[xylar]

conda-forge/conda-forge-repodata-patches-feedstock#279

[saraedum]

Thanks a lot for looking into this @xylar. And true, I had missed the main channel in my case.

Not sure how complicated it will be to get main patched. Maybe the easier solution here would then be to patch/mark broken markupsafe's build 0?

[xylar]

My understanding is that the missing feature that broke junja2 was removed on purpose in markupsafe. So markupsafe isn't broken at all.

[saraedum]

Sorry, I did not mean to say that markupsafe is broken. Rather the build 0 is broken because it is missing a run constraint. Without it jinja on main breaks. Since build 1 and build 0 are identical apart from the metadata, I don't think we need build 0. If we mark it as broken, then it won't be considered anymore and can't break main/jinja2.
Does that make sense?

[saraedum]

(We can also patch the metadata for build 0 so it contains that constraint. I don't think it makes a difference really.)

[xylar]

I see. build 0 didn't have the run_constrained to keep jinja2 < 3 from being installed.

Yes, I think you could get build 0 marked as broken. Or you might be able to get the run_constrained added in https://github.com/conda-forge/conda-forge-repodata-patches-feedstock/. I'm not sure what is possible and what isn't.