Should WCSNAME* be case-insensitive? about stwcs HOT 11 CLOSED

Like all FITS keywords.

from stwcs.

Like all FITS keywords.

No, this is different. This is not about wcsname being the same as WCSNAME but rather:

WCSNAME = SOME NAME

being the same as

WCSNAME = Some Name

or

WCSNAME = some name

from stwcs.

In other words, if we want to have only a single Alt WCS with a given WCSNAME, should the following be allowed or not:

WCSNAMEA=Some Name
WCSNAMEB=SOME NAME
WCSNAMEC=some name

?

from stwcs.

Oh, you are asking about the value. There's no convention.

from stwcs.

Current code treats WCSNAME values as case-insensitive for the purpose of identifying corresponding WCS keys. Should is this how we want to have WCS names be when they are compared?

from stwcs.

There's no convention.

This suggests that the comparison should not be case-insensitive, i.e., name1 should be considered distinct from NAME1, right?

from stwcs.

Treating WCSNAME values as case-insensitive makes it easier for the user to see that there are differences in the WCSs. It would be immensely confusing (as if it isn't already confusing enough) to explain to a user why a WCS named 'IDC_a123458bq-hsc30' is not the same as 'IDC_a123458bq-HSC30' since both reference 'HSC30'. Besides, users are already used to FITS keywords being case-insensitive, so simplifying the treatment of alternate WCSs by making them case-insensitive seems to be consistent with FITS. Therefore, it would be the most user-friendly, in my opinion, to keep WCSNAME as case-insensitive.

from stwcs.

But why would our software sometimes write 'IDC_a123458bq-hsc30' and sometimes 'IDC_a123458bq-HSC30'? I think the solution to this question now should be based on whether our existing software mixes cases when writing WCS names.

from stwcs.

I would not want to try to second-guess how these situations arise. However, users can provide a list of catalog names as an input parameter to the alignment code which would then get used to generate these names. If they provided lower-case one time and upper-case another (interactively, without thinking), they would end up with mixed cases as separate WCSs even though they may be identical. So, there are ways in which the user can cause this situation and make having WCSNAME be treated as case-insensitive be useful.

from stwcs.

Thanks!

from stwcs.

@stsci-hack But then there is code that is case sensitive, like here:

It just seems that there is inconsistency in treating WCS names.

from stwcs.