Use Case 5 : Discussion

[brianthomas]

Comments on Use Case 5 imported from wiki

• I think this could also be expanded to mean the converse: that we have to hold at least the information that is contained in FITS file in the new data format. This is also critical for migration. Or perhaps this a separate use case? =@brianthomas
• I (@timj) don't think this is a use case at all. The above is expressed as a requirement. There are two approaches here: migrate common astronomy tools to support the format natively and, secondly, provide an application (much like Starlink CONVERT) to allow import and export from/to FITS. As @brianthomas says, the requirement on the format itself is that all data currently contained in a FITS file can be represented in the new format (the converse does not have to be true although it does lead to pain and suffering for the people that haven't migrated to the new format yet). Everything else is an implementation detail.
• @timj sounds like you object on 2 counts: wording of the description and the need for this. Based on the ground rules, its not fair to discount the need; you may decide (along with others) to vote down related requirements later. I do think we might ask for more clarification; perhaps the author meant the converse case or has other details which might make this less controversial. Either way, I'll make a note to include the converse case should it not be covered here. =@brianthomas
• Ok, but use cases have to be written as use cases. Requirements derive from use cases. Anything of the form "XXX should/shall happen" is a requirement. I wasn't objecting to the requirement that we provide format conversion facilities. The use case is something along the lines of Astronomer X wants to use common tools such as DS9 to display the data and IDL to analyze it. (@timj)

[sergiopasra]

I have rewritten the use case, it's true that I focused in the requirement and not in the use case. The basic idea is, how will we use tools that do not support the data format. It's true that a lot of programs will be updated to support it, but others (tools without man power or even abandoned) will not.

[embray]

How many "abandoned" tools are out there that are even worth using? Don't get me wrong, I'm sure they exist in vast quantities. For nearly any given piece of software someone out there is using it to do their analysis. But as long as there are "good enough" conversion tools they can keep on doing that.

[sergiopasra]

This is not only for "abandoned tools", but also for tools that lack manpower to do fact updated. At least one fairly used astronomy application falls in the category "large and old codebase, very few maintainers", in my opinion.

[brianthomas]

Just a note I tried to extract requirement 6 from this use case. You may wish to look and comment.

[embray]

Makes sense...

(now I'm having wild-haired imaginings of filesystem plugins that transparently translate new_file_format to FITS to work with older tools...)

[timj]

The NDF library already manages to do behind the scenes format conversion. You give a FITS file to a Starlink application and it will convert it to NDF without you knowing about it.