bug#39293: [PATCH] Base bookmark-bmenu-mode on 'tabulated-list-mode'

[Drew Adams]

> > The attached patch changes bookmark-bmenu-mode to be based on
> > tabulated-list-mode instead of special-mode.
> >
> > This allows us to simplify the code in several cases.  In
> > addition, we get many features for free, such as sorting columns
> > by clicking on the column headers, changing size of columns.
> 
> Both features looks interesting to me. Thanks for working on that!
> 
> > In the future, this will obviously include any new feature added
> > to 'tabulated-list-mode'.
> >
> > The only functional step backwards is that we no longer support
> > the optional "inline" header line -- a bookmark.el-specific hack
> > to have a header without using 'header-line-format'.  I don't
> > believe this feature is very useful since the lack of such
> > support for anything similar in e.g. 'package-menu-mode' has not
> > caused any problems.  It seems to have been added together with
> > 'header-line-format' as a fire escape if the latter caused any
> > problems.

Please don't do that.  As I said earlier:

https://debbugs.gnu.org/cgi/bugreport.cgi?bug=39293#8

this is totally incompatible with the many
enhancements I've made to bookmarks.  It will
force me to include the (former) bookmark.el
code in Bookmark+, and separate the latter
completely from bookmark.el.  Until now, it's
been a compatible extension.

Tabulated-list mode is a step backward for
any feature that offers more.  Just because
something is capable of displaying some info
in columns, that doesn't mean it should be
reduced to the limitations of t-l mode.

Thinking that t-l mode is for all columnar
presentations and interactions is akin to
thinking that a soda straw is appropriate
for an oil pipeline.

This is serious for me.  If you don't care,
too bad for me, and for anyone who cares
about Bookmark+ and compatibility with
vanilla Emacs.

> I am the one who introduced `header-line-format in bookmark.el
> (7a78e19f24) and I confirm that I kept the legacy header "just in
> case".  The new behavior has been enabled by default since 2013,
> and I've not heard any complaint.

I would have complained about it.  But I've
been able to ignore it, because a "legacy
header" is still supported.  Enabled only by
default means that hearing no complaints
signifies nothing.

If a user option is provided to let users NOT
have tabulated-list display imposed, and let
them continue as before, then fine.  Otherwise
I'll be forced to fork, I'm afraid.