[fitsbits] start of Public Comment Period on the INHERIT convention

[Frank Valdes]

Since NOAO is one of the principle proponents and (active) providers of the INHERIT functionality I summarize the points to be considered from my perspective.

The primary operational argument for being part of the standard is to push developers of applications (e.g. sextractor) and libraries (e.g. cfitsio) to support inheritance when INHERIT=T is seen.  This is for the benefit of the user community (as opposed to the developer community) so that they are not left wondering why some tool fails.  This would mean that the standard would have to spell out in some detail what the behavior of inheritance is to be in various cases.
Apart from this argument I would be happy to have this remain a convention with the standard simply reserving INHERIT, briefly summarizing its purpose, and possibly providing a list of keywords that should never be inherited (the convention already gives some of this but it is not up-to-date in regards to something like binary tables).  The purpose of such a section is, again, to help (format) developers know that some set of keywords should be duplicated in all extensions while leaving large volumes of telemetry and documentary keywords available in just the PHU.
Whatever is concluded, it should not convert a currently legal convention into a file that is no longer legal FITS (the once FITS always FITS rule) and, in my opinion, should not provide a value judgement of whether this is good or not.  I don't think anyone has advocated this but I want to be clear.
I am also on the side that discussion of usage and conventions DOES belong in a standards document rather than just being a dry specification of the "standard" legalize.

Regards,
Frank Valdes
NOAO Pipeline Scientist

P.S. I've also refrained from mentioning one of the elephants in the room because there is always such a knee jerk reaction, but IRAF supports INHERIT transparently in all its applications (for runtime FITS) for the benefit of the many users in its community.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/fitsbits/attachments/20150702/605cca68/attachment.html>