Add issue templates #590
Add issue templates #590
Conversation
| required: true | ||
| - type: input | ||
| attributes: | ||
| label: Version |
There was a problem hiding this comment.
Instead of just the TorchGeo version, we could expand this to the full development environment, including:
- OS: e.g., macOS 12.4
- Python: e.g., 3.9.12
- TorchGeo: e.g., 1.2.3.dev0 (e1285e6)
- Installation method: e.g., pip, conda, spack, git, etc.
- Dependency versions: e.g.,
pip list,conda list,spack find, etc.
I'm not sure how useful the rest of these are since TorchGeo is pure-Python and pretty heavily tested...
There was a problem hiding this comment.
Yeah I think torchgeo version is fine for now (i.e. until we see something crazy that is not caught by tests or explained by dependency/Python version)
| Thanks for taking the time to report this bug! TorchGeo is an open-source project | ||
| maintained by its users. If you're Python savvy and want to contribute a pull | ||
| request to fix this bug, we'll be happy to review it. If not, we'll try to fix it | ||
| as long as we can reproduce it. |
There was a problem hiding this comment.
Wanted to clarify that TorchGeo isn't a commercial product with a team of developers just waiting to fix your bugs lol
| blank_issues_enabled: true | ||
| contact_links: | ||
| - name: ❓ Questions | ||
| url: https://github.com/microsoft/torchgeo/discussions |
There was a problem hiding this comment.
Could replace this with a Slack invite link if we ever decide to go down that route
| required: true | ||
| - type: textarea | ||
| attributes: | ||
| label: Steps to reproduce |
There was a problem hiding this comment.
A lot of other repos have sections for "expected result" and "actual result" but I always find myself writing something like "I would expect this to work" and "it doesn't work" so I've never found those sections super useful. Hopefully users will clarify when things don't work as they would expect.
| ... | ||
| ``` | ||
| validations: | ||
| required: true |
There was a problem hiding this comment.
My only concern is making the reproducible example required. I don't want someone to not report something just because they have to provide a reproducible example. Maybe we can explain that bugs are much easier to fix if there is reproducible code and we don't have a lot of free time to make our own examples?
There was a problem hiding this comment.
I think having a reproducible example is a low bar. At the bare minimum, just upload the code you ran and the files you ran it on to produce the error message. Having a minimal reproducible example is a bigger hurdle, but users rarely consider that to be a requirement unfortunately. If someone reported an issue with just an error message and didn't tell me how they got that error message, I would probably just close the issue. I would go so far as to say having a reproducible example is the only thing that is absolutely required to report an issue.
There was a problem hiding this comment.
That's totally fine -- just a minor concern! The more I think about it the more I agree, we really don't have a ton of time to dig into things (esp. if they are obscure).
adamjstewart
force-pushed
the
issues/template
branch
from
4233cd7 to
64a973c
Compare
|
Added a note explaining the scope of features we accept. I think having a reproducible example should absolutely be required. |
* Add issue templates * Explain scope of TorchGeo features
This PR adds issue templates to the repository. They are loosely based on PyTorch's and Spack's issue templates. See the official docs for more details on possible values and syntax.
I don't really know how to view/test these without forking the repo and merging them to main in the fork? I guess we can do all the wordsmithing in this PR and fix any bugs we find after merging.
Closes #584