Re: [RP] new fdump structure may affect scripts

[Shawn Betts]

"Bernhard R. Link" <address@hidden> writes:

> * jesus c. meyendriesch <address@hidden> [041202 22:06]:
> > there is a problem in the frestore function. The cause seems to be
> > strtok. strtok_r is probably better. Therefore is here a small patch for
> > it:
> 
> This fixes also my problems with :undo. Without it only the first frame
> is restored. Could this please be applied?

Yes, when I get a chance to review the patch. This week
sometime. double fer sure.

-Shawn



From a binary distro perspective it must be painful to be stuck with
my favorite hotkey :).

>
>  * vertically or horizonatally split windows, like emacs or IBM 5250 
>    terminals -- but this probably really is bloat, and it's also hard
>    to get the behaviour right.

Oh yes, this will be implemented at some point. 

>
>At any rate, thanks.  I'll buy you a beer / other beverage if you'd
>ever around Canberra.

Well then! A beer it is!


From the screen manpage:

       C-a C-t     (time)        Show system information.

So you can use it :)

G'luck,
Peter

-- 
I am jealous of the first word in this sentence.


From you patch, I gather that you made a copy of the tree, worked in
that and diffed against the snapshot.  Is this true?  Can you not
simply work in the cvs tree and before you post your patch, do another
cvs update to resolve any conflicts?  You can then also use cvs diff
to generate your patch.

Thanks for the patch.  I would like to see most of those changes put
into the source.

What do you think shawn?

Ryan

--
When this load is DONE I think I'll wash it AGAIN..


From the ChangLog:

        * actions.c (key_actions): added additional keybindings to include
        C- versions for all keys.

in events.c in handle_key() to check if the key pressed matched a
bound key, it would check if the modifiers matched the modifiers of
the bound key. But if the modifiers of bound key was -1 then any
modifiers were valid. This seems to have been removed.

        * events.c (handle_key): can no longer over modify keystrokes

Is this the changelog entry notifying the removal?

Why was it removed?


From the news file:

,----
| ** conf.h compile-time customizations moved
| They are now accessible through the def* suite of commands.
| 
| ** Window alignment
| Windows can now be aligned so they stick to the top-left corner of the
| screen, the center of the screen, the bottom-right, etc. Use the `pos'
| command. There is a customizable default setting for each of the three
| 'kinds' of windows: `defwinpos' for normal windows, `deftranspos' for
| transient windows, and `defmaxsizepos' for normal windows with
| maximimum size hints.
| 
| ** Window format string
| The format of the list of windows dumped by the `windows' command is
| now customizable with `defwinfmt'.
| 
| ** Different Window naming schemes
| Windows now have 3 resources that can be used as their default name:
| the WMNAME hint which generally contains transient information such as
| what file is currently open, the res_name from the WMCLASS hint, or
| the res_class the WMCLASS hint. This affects commands like `select'
| which search for a window by name.
`----

I'll be writing the docs on these new commands in a bit. The defwinfmt
command should solve Mark's Window ID problem since it can be exported
with this command in conjunction with the windows command. Jonathan
will probably like `pos' since windows can now be 'aligned'.

Anyway, I haven't tested it very hard so let me know what anomalies
'n' bugs you find! I also haven't lined any of the commands with
existing screen commands. I'd wager some should be renamed. If you
find one that should be renamed let me know.


From looking throgh the source, it looks like a lot of the code needed
to support multiple screens is already in place. I.e. - it asks the
server for the screen count, and deals with them. However, even though
I've had as many as 4 X servers running today, I don't think I can get
one to start on multiple screens to test it :-(.

        <mike
--
Mike Meyer <address@hidden>                     http://www.mired.org/home/mwm/
Q: How do you make the gods laugh?              A: Tell them your plans.


From what I understand Shawn is quite busy studying _and_ working.

I'm sure he'll get to it in time. I don't think you are being ignored
:-)

Regards,
Doug







From before, why does barloc take an int, but all the window
positioning things take a gravity? This isn't a screen issue, as
there's no such command in screen.

If I happen to be working on these things, I just may fix them in
passing.

        Thanx,
        <mike
--
Mike Meyer <address@hidden>                     http://www.mired.org/home/mwm/
Q: How do you make the gods laugh?              A: Tell them your plans.


From my .ratpoisonrc:

alias rxvt exec rxvt
unbind o
bind o rxvt


Effect of <command> o:

/bin/sh -c "rxvt" finished (127)


Any ideas?

Joe 


From that angle, rp should have a different default prefix so that it
is screen compatible :).

Shawn



From your list of desiderata, simple categorization, and ease of
adding and removing items are probably the only features that todl
mode explicitly addresses at present.  Intelligent support for
multiline items and file/dir/program/function association are
planned for.  It should be possible to have a good way of marking
important deadline items too... the way I might think to do it
initially is to just maintain a separate list of deadline or
high-priority items (the system is hypertextual, remember, so this
list would be full of links to the expanded treatment of these
items).  You could probably find a way to screen-splash this
information via the ability to associate functions or programs to
items.  (I hope to also eventually provide an integrated replacement
for the appt package, but I'm not quite there yet.)

So I'd suggest taking a look at this mode -- being aware that there
are some bugs that I haven't been able to fix and that the system is
sort of "under transition" right now -- and seeing if you like the
basic framework.  If so, I may be able to add the key bells and
whistles that you're after.

Joe

PS.  It can export to HTML and there is a user manual so-generated
at

http://www.ma.utexas.edu/~jcorneli/y/todl/user/

But the current system is a bit different from than the one
documented there.


From our conversation I gather that I need to be a little clearer on a
couple of points, so I'm posting some of it to the list.

-trent

------------------------------------------------------------------------

> The main reason I was looking for your patch was to make expose
> work.

If you look at the comments at the start of expose.pl, you'll see that
it will `kinda' work without patching on actions.patch. =20

What you need to do is change this block from if($ret) to if(!$ret).

        if ($ret)
        {
            # Good behavior
            &rp('only');
=09
        #   # OS-X behavior
        #   chomp $ret;
        #   my $id =3D  &xid2num(&frame2xid($ret));
        #   print "'$id'\n";
        #   &rp("frestore $frames_old");
        #   &rp("select $id");
        }
        else
        {
            # User aborted the :fselect, so restore the old frameset.
            &rp("frestore $frames_old");
        }

I haven't tested if this still works, so applying actions.patch is
preferable.

> I mean, when I run the expose script, it always returns me to the
> current frame, regardless of my selection.

With actions.patch, expose can tell if you cancel (C-g) the :fselect,
and restore your old frameset.  Unless you
    * Apply actions.patch; or
    * Change the if statement described above,

expose will think you have canceled EVERY :fselect, and restore EVERY
old frameset.  This sounds like what's happening for you.

> Also we had a rather popular discussion about ratpoison
> in Gentoo-forums.
> http://forums.gentoo.org/viewtopic.php?t=3D220150
>
> So I tried to patch ratpoison and it succeeded, but unfortunately
> still no luck with the script -- it always returns me to the current
> frame.

Expos=E9's behaviour should be as follows:
  1. User calls expose.pl.
  2. Expose saves the current frameset.
  3. Expose puts all windows on-screen at once.
  4. Expose calls ratpoison -c fselect
  5a. User selects a frame (by pressing 1 to 9).
  6a. Expose calls ratpoison -c only, causing chosen window to take up the =
whole screen.

If actions.patch applied, you can also get:
  5b. User cancels fselect (by pressing C-g).
  6b. Expose restores the saved frameset.

------------------------------------------------------------------------


From memory, the current history item 'pointer' wasn't being reset to
the end of the list in certain situations such as when editor_abort()
was called.

Regards,
Doug