chardet + dry-run + doc upgrade

Changelog

@@ -5,14 +5,46 @@ Version Changes for Hypermail
 HYPERMAIL VERSION 3.0.0
 ============================
 
+2026-07-15 Jose Kahan
+	* aclocale.m4 configure.ac
+	Upgraded to 1.17, 2.72, respectively
+	Added return type to main for C compiler test in configure.ac
+
+	* hypermail.c setup.c setup.h file.c print.c print.h lang.h
+	Added a new CLI option, -y, to do a dry-run of hypermail, without
+	generating any archive; it will output to STDOUT the filenames
+	and associated Message-IDs it would have created.
+
+2024-11-08 Jose Kahan
+	* src/string.c
+	Truncate Content-Type and charset field values at first illegal
+	character
+
+	* src/parse.c
+	Improve integration of libchardet for converting mail header values to
+	utf-8 / remove compiler warnings
+
+	This change let's the parset detect the charset of a message when the
+	Content-Type: header is missing the charset field or if a header value
+	is in another charset and it's not encoded using RFC2047.  To do so,
+	the parser calls a function that examines the To:, From: and Subject:
+	header values and, when they don't have valid ASCII or UTF-8 strings,
+	if it detects the charset, it will convert the header value to UTF-8
+	and use that charset for the rest of the message. If it fails to detect
+	it, it will replace the whole header value with a "(invalid string)"
+	string.
+
+	Fixed compiler warnings as reported by pedantic, unused variables
+	flags.  src/parse.c
+
 2024-11-07 Jose Kahan
 	* configure.ac, configure
 	Improve detection of libchardet (system, local dir) and enable
 	it by default
 
 	* pkgconf, libiconv
 	Make pkgconf and libiconv mandatory for compiling hypermail
-	
+
 2024-11-06 Jose Kahan
 	* src/hypermail.c
 	If hypermail was called with its default configuration values,
@@ -189,7 +221,7 @@ HYPERMAIL VERSION 3.0.0
 	be taken into account internally (e.g., for threads).
 
 	* src/{hypermail.h, struct.c}
-	When a message/rfc body part contained only a stored
+	When a message/rfc822 body part contained only a stored
 	attachment (e.g., text/html), even if the attachment
 	was being added to the attachments dir, both the message/rfc822
 	headers and the link to the stored attachment were not being

RELEASE_NOTES

@@ -1,9 +1,11 @@
 Hypermail 3.0.0
 
+* IMPORTANT * IMPORTANT * IMPORTANT * IMPORTANT
+
 If you are migrating to 3.0.0 from 2.4.0, please consult the UPGRADE
 file.
 
-*** NEW FEATURES
+*** NEW FEATURES AND CHANGES
 
 - HTML5
 
@@ -63,11 +65,12 @@ anymore.
 - Switch to PCRE2
 
 Up to now hypermail had been using PCRE behind the scenes of its
-message filtering options (docs/hmrc.html#filters). PCRE being now
-at end of life, and is no longer being actively maintained, we have now
-updated hypermail to support its successor, PCRE2. If you were using any
-of Hypermail's filtering options, we advise you to check out PCRE2's
-doc to see if there have been any changes there that may impact you:
+message filtering options (docs/hmrc.html#filters). PCRE being now at
+end of life, and is no longer being actively maintained, we have now
+updated hypermail to support its successor, PCRE2. If you were using
+any of Hypermail's filtering options, we advise you to check out
+PCRE2's doc to see if there have been any changes there that may
+impact you:
 
   https://www.pcre.org/current/doc/html/pcre2pattern.html
 
@@ -79,24 +82,30 @@ the future.
 
 - Other configuration option changes
 
-Two new configuration options, "archive_date" and "hypermail_colophon",
-allow you choose whether you want to display a line that says when the
-archive was generated (only in the indexes) as well as a a line that
-states that the archive was generated with hypermail and the
-generation date (both in messages and indexes). Both these options may
-be considered mutually exclusive as they both display the archive
-generation date. You can disable both of them.
+Two new configuration options, "archive_date" and
+"hypermail_colophon", allow you choose whether you want to display a
+line that says when the archive was generated (only in the indexes) as
+well as a a line that states that the archive was generated with
+hypermail and the generation date (both in messages and indexes). Both
+these options may be considered mutually exclusive as they both
+display the archive generation date. You can disable both of them.
 
 "empty_archive_notice" will let you customize the markup and text that
 is displayed in indices when you're converting a mbox that consists
-exclusively of messages that have been annotated as 'spam' or 'deleted'.
+exclusively of messages that have been annotated as 'spam' or
+'deleted'.
 
 "show_headers_msg_rfc822" will let you customize the list of headers
 that you want to be shown in message/rfc822 attachments. If not
 defined, hypermail will use the value of "show_headers".
 
-"archived_date" lets you control whether you want to add a line
-in the indices giving the date the archive was generated.
+"archived_date" lets you control whether you want to add a line in the
+indices giving the date the archive was generated.
+
+"ignore_content_disposition" allows you ignore the Content-Disposition
+header for a given list of MIME types. This is particularly useful for
+old Apple Mail UA that associated Content-Type: attachment with
+multipart/appledouble.
 
 Please refer to refer to hmrc.html hmrc.4 or use  hypermail -v for a
 more detailed description of the configuration options.
@@ -113,9 +122,25 @@ to.
 This new parser has allowed us to better handle message/rfc822 and
 multipart/alternative attachments, simplify hypermail's parsing code,
 and use better heuristics for determining the charset that will be
-associated with a processed message. The new parser also greatly
-simplifies adding new markup changes as it allowed to separate the
-parsing from the markup generation.
+associated with a processed message (some of these heuristics use
+libchardet). The new parser also greatly simplifies adding new markup
+changes as it allowed to separate the parsing from the markup
+generation.
+
+- libchardet
+
+Hypermail now uses libchardet (if installed) to help detect a
+message's character set when the message's Content-Type has no
+declared charset, header values are not ASCII, and there are not
+either RFC2047 encoded.
+
+  https://github.com/Joungkyun/libchardet/
+
+- libiconv
+
+Due to the predominance of UTF-8 and the wide availabilty of libiconv,
+we have made this library mandatory, rather than optional, for the
+compilation of hypermail.
 
 - Compiler warnings and memory leaks
 
@@ -126,16 +151,31 @@ Although all major runtime cumulative memory leaks that gcc's
 code that were not tested as they depend on the use and combination of
 different hypermail configuration options.
 
+- Sample configuration file
+
+An annotated configuration file and related files is now
+available in docs/examples/hmrc.conf.
+
 - Other code improvements
 
-Read the Changelog for more details.
+Please consult the Changelog
+
+*** UNMAINTAINED FEATURES, USE AT YOUR RISK
+
+It's been a while since Hypermail's support of gdbm (--with-gdbm) for
+storing message's metadata and libfnv (--enable-libfnv) for creating
+unique filenames have been reviewed and tested.
+
+They're now downgraded to UNMAINTAINED status and may be deprecated in
+a future version unless users report they are using them and that they
+work as expected.
 
 *** DEPRECATED FEATURES
 
-The configuration variable "indextable", which let hypermail
-generate message indexes using tables has been deprecated in favor of
-using HTML and CSS. This option has been disabled.  The pertaining
-code will be removed from the code base in a future release.
+The configuration variable "indextable", which let hypermail generate
+message indexes using tables has been deprecated in favor of using
+HTML and CSS. This option has been disabled.  The pertaining code will
+be removed from the code base in a future release.
 
 The "quotes" and "finequotes" options are now in the to be deprecated
 list. The code that handles these options hasn't been updated or
@@ -184,6 +224,9 @@ those issue trackers to send your feedback.
 - translation of messages: some languagess are behind since years and
   need update. Unmaintained languages risk being dropped out in a
   future version of hypermail. Issue #78 on github.
+  All these messages should also probably be converted to UTF-8 and
+  have hypermail change them on-the-fly and hypermail should require
+  UTF-8 locales.
 
 - compilation: windows LCC support (not updated, seems very
   old). Issue #79 on github.
@@ -203,3 +246,5 @@ facilitating discussions and work on this version of hypermail.
 
 Thanks to Baptiste Daroussin, Jim (@AverageGuy), @cacsar, @schlomif,
 and Andy Valencia (@vandys) for their bug reports and code contributions.
+
+Thanks to @Joungkyun for libchardet and insights into its use and status.