Re: [Nmh-workers] mh-format and components with empty content

[norm]

Ken Hornstein <address@hidden> writes:
>>I  want to write a Java program implementing a heuristic to determine which
>>Emails to backup to external media. The heuristic will depend on such
>>factors as a message's spam score, it's age, and whether or not it is
>>backed up elsewhere in the cloud. Whether I backup a message will depend
>>on its total merit.
>>[...]
>>BUT my question was just a question NOT a request. This is because
>>when "%{Nag}" returns null I can workaround by reading the lines
>>of the message to see which comes first, on the one hand an empty string
>>(indicating that I have read all the headers) or an
>>an end  of file, or on the the other hand the string "Nag:".
>
>So, given what Norm has said ... what do people think?  Implement, or no?

On the one hand, not distinguishing between no string and an empty string is
at best an inelegance and at worst a bug.

On the other hand, as David Levine points out, Emails having headers with
empty content are extremely rare and there are simple workarounds. So unless
the fix is trivial, I would rather that the effort that would go into a fix,
be spent on designing and implementing a complete replacement to mh_format.

In summary, I vote no, unless the fix is trivial.

    Norman Shapiro