[fitsbits] reopening of Public Comment Period on the CONTINUE convention

[William Pence]

On 2/26/2016 12:46 PM, Tom McGlynn (NASA/GSFC Code 660.1) wrote:
> I don't see any significant issues with this proposal.  Tiny nits below.
>     Tom McGlynn
>
>
>    - I don't like the difference in the requirement that there be a 
> space before the / in comments
>      in CONTINUE versus a recommendation that there be such a space in 
> other keywords. 

I agree that it would be more consistent to just recommend that the 
space be present, rather than require it.


>
>   - Given that there is some reference to the historical background of 
> this convention I think it might
>     be appropriate to note that this convention is now to be applied 
> regardless of the existence of
>     a LONGSTRN key in the header.  Maybe the reference in the appendix 
> is enough...
>

I think mentioning this in the appendix is sufficient.

>   - It would be nice if there were some discussion of the value of the 
> comment section when there
>     are multiple comments.  E.g., what is the value of the comment 
> associated with KEY in the following
>
> KEY = 'abc&' ./ Comment1.
> CONTINUE 'def&' /          Ten spaces after /.
> CONTINUE '' / More...
>
> Do I preserve spaces after the /, are they completely insignificant, 
> should I simply put a single space between lines and ignore leading
> and trailing spaces otherwise.  I could imagine:
>    ' Comment1          Ten spaces after /. More ...'
> or
>   'Comment1Ten spaces after/.More ...'
> or
>   'Comment1. Ten spaces after /. More ...'
>
> I think I'd prefer the last.  [Leading and trailing spaces 
> insignificant, but all lines but the first preceded by space.] I 
> wouldn't mandate this but some statement indicating preferred practice 
> would make sense.
>

I agree we should not mandate how spaces are interpreted.  Since the 
comment field is mainly provided for human interpretation and is not 
intended to have a precise string value,  I don't see any need to try to 
prescribe preferred practice because it is too difficult to allow for 
all possible usage.  For example,
one might split a long word with a hyphen to continue it onto the 
comment field of the next keyword record, in which case one would not 
want to insert a leading space.  Or one might use single or double 
quotes in the comment fields to enclose a literal string that extends 
over multiple records within which all the spaces are significant.  I 
see no harm in leaving the interpretation of the comment fields somewhat 
vague.


>    - Not  clear to me what
>
> KEY = 'abc&'
> CONTINUE
>
> is supposed to mean.  Is the CONTINUE keyword conforming so that 
> KEY=abc or non-conforming (no string in it) so the ampersand is 
> significant?  I think I prefer the former, but not obvious what 
> 'conforming' means.
>

The 'CONTINUE' keyword in that example is not a 'conforming CONTINUE 
keyword' because it is not followed by a quoted string, so strictly 
speaking, the '&' at the end of the previous string should be considered 
significant.   'Conforming' means that it satisfies the previous 
definition of the CONTINUE keyword format, which seems pretty 
unambiguous to me.

>    - Do we want to dis/encourage this convention to be used when the 
> string value would fit, but we have a long comment?  E.g.,
> KEY = 'short&' / The start of a very, very, very...
> CONTINUE '&' / very, very, very, ...
> CONTINUE '&' / very, very, very, ...
> CONTINUE '' / long comment
>
> If this is a practice we want to encourage, then the limitation of 
> CONTINUE to after string keywords makes less sense.

I don't see a need to encourage or discourage this usage.  It could be 
useful in cases where the string value almost fills the 80-character 
keyword record and doesn't leave enough room for a useful comment.  In 
the case of keywords with logical, integer, or floating point value, on 
the other hand, there are typically about 50 spaces left for the comment 
field.

-Bill Pence