Re: lynx-dev patch: fix for disappearing empty lines when dump'ing

[Vlad Harchev]

On Wed, 5 Jan 2000, Klaus Weide wrote:

> On Wed, 5 Jan 2000, Vlad Harchev wrote:
> 
> >  Tom, expect a huge offsets applying this patch (I still use pre-dev16, and 
> > it
> > differs from dev17 in so many of places...). I inserted fragments of dev17's
> > print_*_to_fd to dev16's GridText.c, saved as 'old/src/GridText.c', edited 
> > and
> > tested code, and saved as 'fixef/src/GridText.c', and made a diff (So I hope
> > that 'patch' won't bark). 
> >   1st hunk is at the very begining of split_line.
> > 
> > diff -ru old/src/GridText.c fixed/src/GridText.c
> > --- old/src/GridText.c      Sun Jan  9 06:15:29 2000
> > +++ fixed/src/GridText.c    Sun Jan  9 06:26:18 2000
> > @@ -2385,6 +2385,22 @@
> >      cur_hy_word_size = 0;
> >  #endif
> 
> It won't work for patch, there is not cur_hy_anything in dev.17, but
> I'll let Tom try to deal with that...

  Oops. But I hope there won't be any other problems with it. As I said, 1st
hunk is at the begining of split_line.
 
> >  
> > +    cp = previous->data;
> > +    if (cp[0] == LY_BOLD_START_CHAR || cp[0] == LY_UNDERLINE_START_CHAR) {
> > +   switch (cp[1]) {
> > +       case LY_SOFT_NEWLINE:
> > +           cp[1] = cp[0];
> > +           cp[0] = LY_SOFT_NEWLINE;
> > +           break;
> > +       case LY_BOLD_START_CHAR:
> > +       case LY_UNDERLINE_START_CHAR:
> > +           if (cp[2]==LY_SOFT_NEWLINE) {
> > +               cp[2] = cp[1];
> > +               cp[1] = cp[0];
> > +               cp[0] = LY_SOFT_NEWLINE;
> > +           };
> > +   };
> > +    };
> 
> I think this will normally work.  But being at the beginning of split_line,
> it will actually apply with a one line delay.  First split_line() gets called,
> *then* LY_SOFT_NEWLINE gets appended - at least in the "normal" code (I don't
> know whether and hoow this changes with your various options).  The the
> fixup doesn't apply to the LY_SOFT_NEWLINE for *this* line break, but
> for the previous one.  Confusing, and in some obscure situation it may never
> apply (if the last line never gets terminated with split_line()...)
> 
>    Klaus
> 

 Yes, you are right that it will apply fixup with one line delay. End line
 case is not important IMO.
 But current approach requires fixup only in one place, rather in several. And
 as I said, I tested the patch and it works fine.

 Best regards,
  -Vlad