You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
numpy: a bit more work, and quite ubiquitous in our ecosystem's packages in any case. Will look into it, but not a priority at the moment.
xmlschema: @gbrunin I found your comment when cleaning up in b962e96. ^^ I'll look into it, but I think it's a Python-only package that only has one other dependency (https://pypi.org/project/elementpath/ - no dependencies). There is also https://pypi.org/project/qeschema/, which might be looking into to simplify our life even further. There is also https://pypi.org/project/lxml/, but not sure if that does what we need. Alternatively we could also skip schema validation, but I'm not sure that's a good idea. Sticking with xmlschema, a tool that is developed by the QE people, seems like the most reasonable approach at first glance.
packaging: this was apparently introduced in e3f9744 as a way to deal with QE versions. Although I'm also wondering what do to with QE versions (which ones to support, how to deal with differences), this seems a little over-engineered to me. QE versions are not so complicated that they require an extra dependency to deal with them. Besides, even though QE versions look like SemVer, I don't they really adhere to it, and so trying to use PEP 440 to determine the right version might do more harm than good.
As @davidwaroquiers notes here in this comment, we currently have several dependencies that could potentially be removed:
scipy📦 Removescipydependency #96numpy: a bit more work, and quite ubiquitous in our ecosystem's packages in any case. Will look into it, but not a priority at the moment.xmlschema: @gbrunin I found your comment when cleaning up in b962e96. ^^ I'll look into it, but I think it's a Python-only package that only has one other dependency (https://pypi.org/project/elementpath/ - no dependencies). There is also https://pypi.org/project/qeschema/, which might be looking into to simplify our life even further. There is also https://pypi.org/project/lxml/, but not sure if that does what we need. Alternatively we could also skip schema validation, but I'm not sure that's a good idea. Sticking withxmlschema, a tool that is developed by the QE people, seems like the most reasonable approach at first glance.packaging: this was apparently introduced in e3f9744 as a way to deal with QE versions. Although I'm also wondering what do to with QE versions (which ones to support, how to deal with differences), this seems a little over-engineered to me. QE versions are not so complicated that they require an extra dependency to deal with them. Besides, even though QE versions look like SemVer, I don't they really adhere to it, and so trying to use PEP 440 to determine the right version might do more harm than good.