branch master updated: * tp/Texinfo/Convert/ParagraphNonXS.pm (new): rename 'last_char' as 'last_letter'. a Always set 'last_letter', to an empty string when it was previously undef.

[Patrice Dumas]

This is an automated email from the git hooks/post-receive script.

pertusus pushed a commit to branch master
in repository texinfo.

The following commit(s) were added to refs/heads/master by this push:
     new 19bc99b9a4 * tp/Texinfo/Convert/ParagraphNonXS.pm (new): rename 
'last_char' as 'last_letter'. a Always set 'last_letter', to an empty string 
when it was previously undef.
19bc99b9a4 is described below

commit 19bc99b9a4b2e21c4c4eb913fdaee86bc569a8a4
Author: Patrice Dumas <pertusus@free.fr>
AuthorDate: Tue Jul 25 16:12:11 2023 +0200

    * tp/Texinfo/Convert/ParagraphNonXS.pm (new): rename 'last_char'
    as 'last_letter'. a Always set 'last_letter', to an empty string
    when it was previously undef.
    
    * tp/Texinfo/Convert/ParagraphNonXS.pm (end, _add_next, add_text),
    tp/Texinfo/XS/xspara.c (xspara__end_line, xspara_end)
    (xspara_add_text): set last_letter to the last character when it is
    space or end of line, or fullwidth character.  Do not unset
    last_letter in _add_pending_word.  Unset last_letter in end.
    
    * tp/Texinfo/Convert/ParagraphNonXS.pm ($end_sentence_characters):
    rename $end_sentence_character as $end_sentence_characters.
    
    * tp/Texinfo/XS/xspara.c (after_punctuation_characters)
    (end_sentence_characters): use defin for the strings to be sure to
    avoid errors in code (] was missing in one place).
    
    * tp/Texinfo/Convert/ParagraphNonXS.pm: try to have all debug messages
    as one string.
    
    * tp/t/paragraph.t: more tests with fullwidth characters, including
    latin fullwidth characters that can be upper case.
    
    * tp/Makefile.tres, tp/t/plaintext_tests.t
    (split_punctuation_detection_in_commands): new test with all the
    puctuation related character tested and with @-commands.
---
 tp/TODO | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/tp/TODO b/tp/TODO
index 030dc7e9c2..914d29a001 100644
--- a/tp/TODO
+++ b/tp/TODO
@@ -12,6 +12,8 @@ Before next release
 
 comment on linemacro call at top level comments until end of line.
 
+add @nodedescriptionblock command (if no more input from the mailing list)..
+
 Bugs
 ====
 
@@ -48,6 +50,18 @@ For converter writers,
 Delayed bugs
 ============
 
+For Paragraph, not bugs but improvements and synchronizations of
+the two implementations.  It would be nice to change the XS code
+to operate on multiple spans of characters at once when relevant,
+we'd have an extra step of keeping track of how long each span of
+text is, and then processing that span of text together (in the loop
+in xspara_add_text).  It could potentiall have performance benefits
+(e.g. reducing the number of function calls) and be easier to debug,
+with fewer steps in the processing to trace.  The XS code is easier
+to read for the ens of sentence handling part, so for that part, it
+could be relevant to use the XS code as reference and change the perl
+Paragraph.
+
 See message/thread from Reißner Ernst: Feature request: api docs
 
 hyphenation: should only appear in toplevel.