[tin-dev] [tin 2.4.3] snapshots - please test

[Urs Janßen]

In <mailman.191.1531218476.592.tin-dev at tin.org>, Corinna Vinschen wrote:
>> untested (I ommited the configure diff, so you either need to
>> run autoconf (special version needed
>> [<ftp://ftp.invisible-island.net/autoconf/autoconf-2.13-20110430.tgz>])
>> once or hard define HAVE_UNICODE_UNORM2_H for the time beeing):

last patch lacked "include <unicode/unorm2.h>"

--- include/tin.h	2018-06-04 11:54:32 +0000
+++ include/tin.h	2018-07-11 05:13:04 +0000
@@ -1406,9 +1406,13 @@
 #	ifdef HAVE_UNICODE_USTRING_H
 #		include <unicode/ustring.h>
 #	endif /* HAVE_UNICODE_USTRING_H */
-#	ifdef HAVE_UNICODE_UNORM_H
-#		include <unicode/unorm.h>
-#	endif /* HAVE_UNICODE_UNORM_H */
+#	ifdef HAVE_UNICODE_UNORM2_H
+#		include <unicode/unorm2.h>
+#	else
+#		ifdef HAVE_UNICODE_UNORM_H
+#			include <unicode/unorm.h>
+#		endif /* HAVE_UNICODE_UNORM_H */
+#	endif /* HAVE_UNICODE_UNORM2_H */
 #	ifdef HAVE_UNICODE_UIDNA_H
 #		include <unicode/uidna.h>
 #	endif /* HAVE_UNICODE_UIDNA_H */


> linked against the latest libicu-62.  Testing with english, spanish and
> chinese UTF-8 news articles looks good to me.

normalize() is only called for search actions, just viewing articles
doesn't involve it.

searching for U+212B (ANGSTROM SIGN) in a text containing U+00C5
(LATIN CAPITAL LETTER A WITH RING) or the like is needed to test it
(with different settings of tinrc.normalization_form).