Case-sensitivity of license IDs in download

[mxmehl]

The SPDX-License-Identifier given as a parameter for reuse download is case-sensitive. I wonder whether this is actually necessary since there is no conflict. Especially for manual downloads, some users might appreciate the tool being less strict about this.

However, I think that we should stay case-sensitive when checking the identifiers with reuse lint.

[carmenbianca]

Quite doable if I use a case-insensitive dictionary (I think requests has one). Requires a bit of fiddling though, so not going to touch it yet because of the low priority.

[mxmehl]

As far as I can see, the IDs are all upper-case. Can the tool just transpose all characters for the ID to upper for this specific command?

However, on the other hand, we should perhaps continue to be strict about this, otherwise people may start using lower/mixed-case chars with addheader. So perhaps rather a warning or suggestion?

[carmenbianca]

the IDs are all upper-case

They aren't. Apache-2.0 is a good example.

However, on the other hand, we should perhaps continue to be strict about this, otherwise people may start using lower/mixed-case chars with addheader

addheader does not verify the user's input at all. Perhaps it should? It might also be possible to implement the same functionality for addheader, e.g., addheader --license Gpl-3.0-or-later would get auto-corrected. I'm not sure how desireable that is, though.

[mxmehl]

Oh, you are so right 🤦‍♂️

I agree that correcting the user is not really desireable. So perhaps an extra information to the error message of "Error: Failed to download license." along the lines of "Please notice that SPDX License Identifiers are case-sensitive" would suffice?

[nicorikken]

I've found some details on this issue from a SPDX discussion thread:

• Clarify case sensitivity of Short Form licenses - for list and tools. spdx/spdx-spec#63
• https://github.com/spdx/spdx-spec/pull/228/files

Case sensitivity is important for tools to operate properly, so producing the correct header is desired. But for human input and matching this is not needed, as case shouldn't matter.

So with that background, I'd like REUSE to autocorrect (only case-wise) the ids that are input by humans.