Re: Duplicate In-Reply-To entries in reply buffer

[Kazuhiro Ito <kzhr@d1.dion.ne.jp>]

I am still confused somewhat, so I wrote notes for my (hopefully our)
understanding.

This problem is related with some bugs and risky codes.
Actual bugs are;

1-1. elmo-msgdb-get-message-id-from-buffer assumes buffer is narrowed
to a message header, but actually it is called from unnarrowed buffer
in some places.

1-2. We have multiple extracting methods for Message-ID and they can't
extract from some valid header.

Risky codes are;

2-1. elmo-field-body is an alias for std11-fetch-field.  FLIM has
std11-field-body function which does not assume buffer is narrowed.
The name of elmo-field-body is confusing.

2-2. The name of elmo-msgdb-get-message-id-from-buffer is confusing.
It makes us to imagine this function extracts Message-ID from whole
buffer, not narrowed region.


I think it would be better we fix 2-1 and 2-2 at first.

Fix (my opinion);

2-1. Create elmo-fetch-field by the same definition of elmo-field-body
and obsolete elmo-field-body.  If there are misused elmo-field-body,
replace them with std11-field-body.

2-2. Change elmo-msgdb-get-message-id-from-buffer not to assume buffer
is narrowed.  If we need the same functionality of the current
elmo-msgdb-get-message-id-from-buffer, create new function,
e.g. elmo-msgdb-get-message-id-from-header.

1-1. Fixed by 2-2's fix.

1-2. David's refactoring is efficient.  As far as I tested, no major
performance problem was observed except using lexical analyzer.  So,
if I can customize to disable lexical analyzer in this part, the most
strict method (such as the combination of lexical analyzer and strict
regexp in [wl-en: 05189]) would be permissive.

-- 
Kazuhiro Ito