[GRASS5] Re: [GRASSLIST:1863] Re: problems displaying layers on 5.7

GRASS GRASS developer
1

(cc grass5)

On Tue, Nov 25, 2003 at 12:24:29PM +0000, Glynn Clements wrote:

Markus Neteler wrote:

[... WISH/GRASS_WISH issues ...]

I have updated etc/Init.sh (lib.init/init.sh) as suggested in CVS.

Following for the grass5 archive and/or tcl/tk programmers:

> > In any case, the ultimate solution would be to fix the portability
> > issues so that using the native wish program works. There really
> > aren't that many portability problems, and some of the ones which
> > exist are gratuitous (e.g. using xwininfo).
> >
> > Regarding the "console1" problems, I was involved in a discussion on
> > the list with Jeshua Lacock back in April, in the thread entitled
> > "More Mac OS X Tcltkgrass Tweaks". However, that thread ended with me
> > asking him to try some more tests; AFAICT, we haven't heard anything
> > from him since then.
>
> http://search.gmane.org/search.php?query=more+mac+os+x+tcltkgrass+tweaks+console1&email=&group=gmane.comp.gis.grass.devel&sort=date
>
> I have remote access without display only (due to firewall or
> whatever restrictions), but I can to a limited extend ask the
> owner of that Mac machine to test something.

OK. AFAICT, that thread contains everything which is publicly known
about the "console1" problems with using the MacOSX Tcl/Tk for
tcltkgrass.

FWIW, the other common tcltkgrass-on-mac problems were:

This does not affect 5.7, right (doesn't contain tcltkgrass)?

1. Use of xwininfo in conjunction with "winfo id".
2. Use of xlsclients and xwininfo in search_xdrivers (gui.tcl).
3. Use of xterm.

1 will never work with the MacOSX Tcl/Tk; the Tk windows aren't X
windows, so you can use xwininfo on them.

2 is valid (XDRIVER is still an X11 program even on MacOSX), but ugly.
If it wants to query the size, it should probably be using e.g.
d.frame.

As for 3, hard-coding "xterm" is gratuitous. Making "xterm" the
default is reasonable, but it should support the option of using a
different terminal emulator. Although getting it to use a "native"
console might be awkward.

--
Glynn Clements <glynn.clements@virgin.net>

[ xterm found in src/tcltkgrass/main/*.tcl ]

Should we introduce
$GRASS_TERM
in etc/Init.sh? However, this is 5.0/5.3 related.

Markus

2

Markus Neteler wrote:

> OK. AFAICT, that thread contains everything which is publicly known
> about the "console1" problems with using the MacOSX Tcl/Tk for
> tcltkgrass.
>
> FWIW, the other common tcltkgrass-on-mac problems were:

This does not affect 5.7, right (doesn't contain tcltkgrass)?

I don't know; I haven't really been following 5.7.

> 1. Use of xwininfo in conjunction with "winfo id".
> 2. Use of xlsclients and xwininfo in search_xdrivers (gui.tcl).
> 3. Use of xterm.
>
> 1 will never work with the MacOSX Tcl/Tk; the Tk windows aren't X
> windows, so you can use xwininfo on them.
>
> 2 is valid (XDRIVER is still an X11 program even on MacOSX), but ugly.
> If it wants to query the size, it should probably be using e.g.
> d.frame.
>
> As for 3, hard-coding "xterm" is gratuitous. Making "xterm" the
> default is reasonable, but it should support the option of using a
> different terminal emulator. Although getting it to use a "native"
> console might be awkward.
>
> --
> Glynn Clements <glynn.clements@virgin.net>

[ xterm found in src/tcltkgrass/main/*.tcl ]

Should we introduce
$GRASS_TERM
in etc/Init.sh? However, this is 5.0/5.3 related.

It's more complex than that, as you typically need to specify both the
command and some of its arguments, and the set of valid arguments
depends upon the program. Also, the position of any variables within
the command line may depend upon the program.

Certainly, each usage pattern should be isolated in a separate
function. E.g. instead of this:

    "Select" "" {
        X0 "" {"exec xterm -iconic -e d.mon select=x0"}
        X1 "" {"exec xterm -iconic -e d.mon select=x1"}
  ...

(from menu.tcl), you would have e.g.:

    "Select" "" {
        X0 "" {"start_mon x0"}
        X1 "" {"start_mon x1"}
  ...

with:
  proc start_mon {name} {
    exec xterm -iconic -e d.mon select=$name
  }

so that any changes would only need to be made in one place.

An easy mechanism for customisation would be to allow functions such
as start_mon to be put into a ~/.tcltkgrassrc file (distinct from the
~/.tcltkgrass file to which tcltkgrass saves its settings) which would
be sourced at start-up.

Allowing the user to override entire functions is both more flexible
and simpler than just allowing them to set variables which are used
within fixed code.

--
Glynn Clements <glynn.clements@virgin.net>