I noticed that in multiqc @ESCAPE_IDENTIFIER@ wasn't being replaced.
It's surprising that this didn't cause any issues on python 2,
but might have to do with the iteration order changes.
This will still write the conda activate part out multiple times,
but fix the check that would otherwise lead to repeated activation
of the conda env. Fixes the slowness reported in https://github.com/galaxyproject/galaxy/issues/7849
Reported by @curtisecombsjr
Fix the following traceback when installing https://toolshed.g2.bx.psu.edu/view/iuc/package_openssl_1_0/ :
```
File "lib/tool_shed/galaxy_install/install_manager.py", line 121, in install_and_build_package_via_fabric
tool_dependency = self.install_and_build_package(install_environment, tool_dependency, actions_dict)
File "lib/tool_shed/galaxy_install/install_manager.py", line 79, in install_and_build_package
initial_download=True)
File "lib/tool_shed/galaxy_install/tool_dependencies/recipe/recipe_manager.py", line 31, in execute_step
initial_download=initial_download)
File "lib/tool_shed/galaxy_install/tool_dependencies/recipe/step_handler.py", line 548, in execute_step
dir = self.url_download(work_dir, downloaded_filename, url, extract=True, checksums=checksums)
File "lib/tool_shed/galaxy_install/tool_dependencies/recipe/step_handler.py", line 77, in url_download
extraction_path = archive.extract(install_dir)
File "lib/galaxy/util/compression_utils.py", line 111, in extract
self.archive.extractall(extraction_path, members=self.safemembers())
File "/usr/lib64/python2.7/tarfile.py", line 2041, in extractall
for tarinfo in members:
File "lib/galaxy/util/compression_utils.py", line 135, in safemembers
raise Exception(finfo.name + " is blocked.")
openssl-1.0.2d/apps/md4.c is blocked.
```
If the user is not an admin.
Fixes https://github.com/galaxyproject/galaxy/issues/7710
An alternative would be to skip Accessibility / Ownership
exceptions, but given that we can just restrict the query
this seems like the better approach to me.
Adjust library folder api to return ldda_id on ld items
and use it for the obect passed to colelction creater
instead of the ld_id.
This bug would only materialize if you have Galaxy isntance that
has a library dataset with multiple versions. Which is the reason why
it eluded us for some time. That and missing tests.
We should only finalize collections up until the map over depth,
as indicated by the TODO that is now obsolete.
Fixes https://github.com/galaxyproject/galaxy/issues/5867 and probably
a bunch of other issues where workflows don't run to completion.
This obviously doesn't work (yet) with filters that work
on the input dataset, but this already removes a lot of
outputs that arent' going to be produced based on static
options. Since exceptions default to priduing the dataset
this isn't much harm.
Nicola Soranzo